26.04.2015 Views

BPM WP FileNET

BPM WP FileNET

BPM WP FileNET

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

A Delphi Group White Paper<br />

<strong>BPM</strong>2003<br />

Market Milestone Report<br />

Contents<br />

Methodology ............................ 2<br />

Defining <strong>BPM</strong> ........................... 4-6<br />

Process Orchestration ............ 7<br />

Phases of Orchestration ......... 8<br />

Orchestration &<br />

Virtualization ........................... 9<br />

Orchestration Summary ........ 10<br />

Components of <strong>BPM</strong> ............... 11<br />

What’s Missing From<br />

Current Offerings .................... 12<br />

Product Selection Criteria ..... 13<br />

Defining Business Logic......... 14<br />

<strong>BPM</strong> Benefits ............................ 15<br />

Integration Priorities ............. 16<br />

Requirements for<br />

Software Suppliers .................. 17<br />

Current State of<br />

<strong>BPM</strong> Deployments ................... 18<br />

Investment Size &<br />

Pricing Models ......................... 19<br />

<strong>BPM</strong> Market Size &<br />

Momentum ............................... 20<br />

Platform Expectations ............ 21<br />

<strong>BPM</strong> Standards ........................ 22-23<br />

Fuego<br />

Product Assessment ................ 24-27<br />

Tesoro<br />

Case Study ................................ 28-30<br />

Fuego, Inc.<br />

2400 Dallas Parkway<br />

Plano, TX 75093, USA<br />

P (972) 801-4200<br />

F (972) 801-4201<br />

www.Fuego.com<br />

Featuring a Delphi Group<br />

Assessment of:<br />

Enabling the New Mission for IT<br />

Today's IT challenges are defined as much by what is in place already as<br />

by what might be missing. We have lived through the "killer app" era,<br />

having spent the last two decades building islands of automation with<br />

packaged software. But the legacy left by these systems is an integration<br />

problem that today consumes the lion's share most IT budgets.<br />

While the challenge of integration remains largely unsolved, it has<br />

nonetheless moved further to the forefront after years of custom<br />

application development, a rising tide of regulatory pressure (now<br />

presenting criminal liability thanks to the Sarbanes-Oxley Act), and for<br />

many firms a morass of system complexity after waves of M&A activity.<br />

Forced to contend with these new challenges in the face of IT dollars<br />

being increasing consumed by integration and maintenance costs, it is<br />

of no surprise that Business Process Management has seen such a rapid<br />

rise in interest over the past year. Yet the real value of <strong>BPM</strong> is not simply<br />

that of a cheaper means for integration, but rather as a resource to<br />

channel existing automation into orchestrated business processes.<br />

Today we are also on the cusp of what may very well be the final great<br />

boom of the software industry, and the last frontier for generating value<br />

from IT infrastructure. Born from the need to keep pace with a<br />

continuously changing business environment, this phase will be led by a<br />

new mandate for software virtualization, componentization, and<br />

orchestration.<br />

If this is starting to sound like Web services, it should. But to lay this<br />

entire opportunity at the feet of Web services would miss the point<br />

entirely. Web services are an enabler, not an end-state. Simply exposing<br />

applications as services will take us little beyond where we are today.<br />

As is illustrated in this report, the true linchpin for the realization of<br />

this opportunity is the ability to connect these components through a<br />

framework of orchestrated business processes --- to transform automation<br />

infrastructure into reusable business assets. This is the new charter for<br />

the partnership of business and IT, and the real value offered by today’s<br />

generation of <strong>BPM</strong> software.


2<br />

Methodology<br />

This report is based on a combination of a recent<br />

Delphi survey and additional analysis by Delphi<br />

Group's research team.<br />

The charts in this report present the findings from<br />

the Q2’03 survey with responses from 500<br />

individual users and evaluators within firms<br />

across multiple industries worldwide.<br />

Approximately 75% of responses are from the U.S.<br />

Respondents varied in their roles from current<br />

users of <strong>BPM</strong> to project sponsors and business<br />

analysts. Approximately 75% of respondents are<br />

presently engaged in either deploying or<br />

evaluating <strong>BPM</strong> software.<br />

The Q2’03 survey follows a Q4’01 research<br />

initiative executed for the 2002 <strong>BPM</strong> Market<br />

Milestone Report. At specific points in this year’s<br />

survey, questions are intentionally repeated from<br />

the Q4’01 survey to demonstrate and validate the<br />

evolving opinions, attitudes and behaviors within<br />

the <strong>BPM</strong> market. Where this was done, charts and<br />

data are presented with both years’ findings<br />

juxtaposed to demonstrate contrast.<br />

Throughout this report, the survey findings are<br />

presented to illustrate concepts and trends, in<br />

addition to the quantified results represented<br />

within the charts. As definitions and perspectives<br />

vary between individuals and across organizations,<br />

examining the contrast in responses offers clearer<br />

reflection of the true state of the market than any<br />

single statistic alone.<br />

Sponsor Initiative(s)<br />

(15%)<br />

Evaluate <strong>BPM</strong><br />

Solution<br />

(6%)<br />

Identify<br />

Requirements<br />

(13%)<br />

Sponsor<br />

<strong>BPM</strong><br />

Initiative(s)<br />

(19%)<br />

Current Role With<br />

<strong>BPM</strong> Deployments<br />

Develop <strong>BPM</strong><br />

Software (3%)<br />

Acquire <strong>BPM</strong> Solution<br />

(Not Sponsor) (9%)<br />

<strong>BPM</strong><br />

End User<br />

(44%)<br />

Lead or Participate<br />

on Deployment Team<br />

(10%) None (3%)<br />

None (10%)<br />

Future Role With<br />

Future <strong>BPM</strong> Deployments<br />

End User<br />

(19%)<br />

In the final stage of this research initiative, Delphi<br />

Group’s research interviewed a set of <strong>BPM</strong> software<br />

vendors and their customers. The results of these<br />

interviews have shaped the analysis presented in<br />

this report, and are also presented at the end as a<br />

set of individual case studies and product profiles.<br />

Define IT<br />

Specifications<br />

(23%)<br />

Define<br />

Business<br />

Requirements<br />

(25%)<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


3<br />

Report Highlights<br />

<strong>BPM</strong> = Workflow + EAI 4<br />

The Anatomy of<br />

Process Management 4<br />

Orchestration vs. Automation 7<br />

Web Services and<br />

Process Orchestration 7<br />

Process Orchestration<br />

Myths and Realities 10<br />

Whole-Product Requirements 11<br />

Market’s Perspective 12<br />

Product Selection Criteria 13<br />

Business Software for a<br />

Business Audience 14<br />

What’s in it for IT? 15<br />

Integration Priorities:<br />

The Market’s Perspective 16<br />

Who Gets The Deal? 17<br />

The Market Comes of Age 18<br />

How Big a Deal? 19<br />

How Big is the Market? 20<br />

Deployment Platforms for Current<br />

and Future <strong>BPM</strong> Initiatives 21<br />

The Role of Standards in <strong>BPM</strong> 22<br />

Product Assessment: Fuego 24<br />

Case Study:<br />

Tesoro 28<br />

Table of Charts & Figures<br />

Current Role With <strong>BPM</strong> Deployments 2<br />

Future Role in With Future<br />

<strong>BPM</strong> Deployments 2<br />

Current Definitions of <strong>BPM</strong> 4<br />

Figure 1 - Conceptual Model of <strong>BPM</strong> 6<br />

Figure 2 - An Automated Process 9<br />

Figure 3 - An Orchestrated Process 9<br />

Process Management Software Landscape 10<br />

Necessary Components of Any <strong>BPM</strong> Solution 11<br />

Features Missing From <strong>BPM</strong> Solutions 12<br />

Overall Functional Priorities Driving<br />

Product Selection Criteria 13<br />

Project Sponsor for <strong>BPM</strong> Investments 14<br />

Responsibility for Defining Business<br />

Rules & Process Logic 14<br />

Single Greatest Functional Advantage<br />

Anticipated from <strong>BPM</strong> 15<br />

Single Greatest IT-Specific Benefit<br />

Anticipated from <strong>BPM</strong> 15<br />

Integration-Specific Priorities<br />

Driving Product Selection Criteria 16<br />

Direction of Increasing Priority 16<br />

Financial Criteria for <strong>BPM</strong> Software Suppliers 17<br />

Investment Timing for Firms Without<br />

Current Deployments or Initiatives 18<br />

State of <strong>BPM</strong> Deployments 18<br />

Reasons Cited for Not Investing in <strong>BPM</strong> 18<br />

Preferred Pricing Model for <strong>BPM</strong> Investments 19<br />

Total <strong>BPM</strong> Software-Only Spending<br />

Anticipated Over the Next Three Years 19<br />

Financial Role for <strong>BPM</strong> Deployments 19<br />

Total Anticipated 2003 Spending on <strong>BPM</strong> 20<br />

Services/Hardware vs. Software 20<br />

Anticipated Platform for Future<br />

<strong>BPM</strong> Deployments 21<br />

App Server Choices for Future<br />

<strong>BPM</strong> Deployments 21<br />

Recognition of Existing <strong>BPM</strong><br />

Standard Initiatives 22<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Defining <strong>BPM</strong><br />

4<br />

<strong>BPM</strong> = Workflow + EAI<br />

Nearly half of respondents to the 2003 survey<br />

identified most closely with the definition of <strong>BPM</strong><br />

as an emerging software layer for building processbased<br />

applications. This is an increase over 2001<br />

and represents a better than 5-to-1 advantage<br />

over any other answer.<br />

This reflects the fundamental shift towards crossfunctional<br />

processes (i.e., across multiple<br />

applications). After spending two decades building<br />

“islands of automation” with packaged software,<br />

organizations today are looking to bridge the gaps<br />

these have left across operational processes.<br />

The growing trend to address this has been greatly<br />

influenced (and vice versa) by changes in software<br />

development that have led to easier separation<br />

of business processes from applications, fueled<br />

in no small part by the Internet and Web services.<br />

In the same manner that the emergence of SQL<br />

and the two-tier client/server architecture enabled<br />

the abstraction of data management from<br />

application code, the emergence of N-tier<br />

computing and component-oriented environments<br />

(such as COM and J2EE) allow for the abstraction<br />

of business processes from application logic.<br />

By separating business process management as<br />

an independent function, applications can be<br />

designed around existing processes, and thus to<br />

take advantage of shared business logic rather<br />

than reinventing and recoding it for each<br />

application.<br />

An emerging layer of<br />

software for building new<br />

process-based applications<br />

Just rebranding<br />

of old-style workflow<br />

A subgroup of EAI software<br />

The latest buzzword<br />

about to be dominated<br />

by Microsoft, IBM, SAP, et al<br />

Have no strong opinions<br />

too early to tell<br />

©2003 Delphi Group<br />

Current Definitions of <strong>BPM</strong><br />

12.3%<br />

19.4%<br />

11.7%<br />

12.1%<br />

4.4%<br />

4.8%<br />

12.3%<br />

9.7%<br />

59.2%<br />

54.0%<br />

2003 Survey<br />

2001 Survey<br />

0% 10% 20% 30% 40% 50% 60%<br />

The notion of cross-functional processes<br />

inevitably brings to mind application integration,<br />

which is no doubt responsible for much of the<br />

confusion between <strong>BPM</strong> and EAI. With these<br />

results, however, the market shows a clear and<br />

appropriate bias towards the independence of <strong>BPM</strong><br />

from EAI or data-centric integration software.<br />

While the market seems clear on this distinction,<br />

<strong>BPM</strong> has nonetheless been pigeon-holed by many<br />

analyst firms as merely a subgroup of EAI. To do<br />

so, however, is to miss the point of <strong>BPM</strong>’s<br />

fundamental value to the enterprise.<br />

<strong>BPM</strong> is not about linking applications, per se, nor<br />

about defining the transformation logic involved<br />

in extracting data from applications. This may be<br />

a subset of what a <strong>BPM</strong> solution offers, yet the real<br />

value of <strong>BPM</strong> is the ability to define and execute<br />

business processes independent of applications<br />

and infrastructure. While EAI and integration<br />

capabilities offer an important resource to <strong>BPM</strong><br />

environments, integration software typically lacks<br />

the ability to address the user-facing side of<br />

business processes. This is consistent with the fact<br />

that EAI is concerned foremost with app-to-app<br />

exchange of data, not user activity or interaction.<br />

The Anatomy of Process Management<br />

To make sense of how <strong>BPM</strong> fits with organizations<br />

and infrastructure, it is helpful to examine the<br />

individual components of <strong>BPM</strong>. While commercial<br />

systems vary in their specific definitions and<br />

software composition, most fit within the basic<br />

framework described below. Note that although<br />

some systems (ERP and HRMS for<br />

example) offer the ability to automate<br />

activities and define business rules, those<br />

that lack the fundamental components<br />

below cannot realistically be used as a <strong>BPM</strong><br />

system.<br />

A <strong>BPM</strong> system is defined by the<br />

components of an Execution Engine,<br />

Process Designer, Process Definitions, an<br />

Activity Monitor, and user interface which<br />

may be a combination of Windows client<br />

application, HTML-based Work Portal, or<br />

an exposed API or Web service (see the<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Defining <strong>BPM</strong><br />

5<br />

model on page 6 for an illustration of how these<br />

pieces work together).<br />

For the vast majority of <strong>BPM</strong> systems on the<br />

market today, application architectures are<br />

component-oriented and allow each of the<br />

individual pieces listed above to be deployed<br />

independently on individual servers, or (albeit<br />

less commonly) as COM or EJB components.<br />

Individual business processes are defined in a<br />

Process Definition and increasingly expressed in<br />

some variation of XML (additional details on XML<br />

standards for Process Definitions are provided in<br />

the ‘Standards’ section of this report, pages 22-23).<br />

Each Process Definition may be composed of both<br />

Manual Activities and Automated Activities. Once<br />

defined and validated within the Process Designer,<br />

Processes are instantiated by an Execution Engine.<br />

An instantiated Process Definition is a Process<br />

Instance. The state, status and history of each<br />

Process Instance are persisted into an external<br />

data store (this is true today of any commercially<br />

viable <strong>BPM</strong> system as none use internal databases<br />

as early systems once did). The Activity Monitor<br />

provides access to status and performance metrics<br />

on the execution of Process Instances. While all<br />

systems provide some form of an Activity Monitor<br />

or reporting mechanism, they vary according to<br />

the degree of control over granularity of reports<br />

and whether or not “live” data can be accessed.<br />

The Execution Engine controls the state and<br />

execution of each Process Instance, which are<br />

composed of multiple Activity Instances. Activity<br />

Instances represent the process activities (also<br />

called Tasks or Steps) during execution, and<br />

include both Work Items and Invoked Applications.<br />

Invoked Applications are automated<br />

activities involving some form of a system-tosystem<br />

transaction, such as a stored procedure<br />

or a Web service.<br />

Work Items are manual activities distributed<br />

to human users or process participants. These<br />

may be a process exception that requires<br />

manual intervention, or a process step that by<br />

definition always requires human input.<br />

When a process instance is executed, Work Items<br />

are generated in response to activities and process<br />

rules, and distributed by the Execution Engine into<br />

an appropriate user’s Work List. The Work List is a<br />

collection of Work Items from one or more process<br />

instances, each with its own state and status. Work<br />

lists represent the integration between process<br />

participants and process instances. Process<br />

participants access Work Items from the work list,<br />

which is coordinated by the Execution Engine.<br />

In this way (i.e., managing the exchange of Work<br />

Items between users and systems) process<br />

management is undeniably similar in appearance<br />

to message queuing and other asynchronous<br />

approaches to application integration (EAI).<br />

Yet it should be understood that managing<br />

processes is not about data flow or moving data<br />

from one bucket to another, nor responding to<br />

discrete events in isolation, but managing flow<br />

control across an entire business process instance.<br />

The ability to include manual activities as part<br />

of the process (i.e., rather than simply part of an<br />

error handling routine) is a subtle but significant<br />

point of differentiation between <strong>BPM</strong> and EAI.<br />

Generating Work Items and managing Work Lists<br />

requires the Execution Engine to have access to<br />

user identities (i.e., through Directory Services)<br />

and access privileges. Roles and interrelationships<br />

(such as team structures) can be defined and<br />

managed within the <strong>BPM</strong> system, but architecture<br />

of most systems on the market today leverage<br />

authentication and directory services from<br />

existing infrastructure using standard interfaces<br />

such as LDAP, JNDI and Active Directory.<br />

Handling Work Items is a critical role of the<br />

Execution Engine and underscores the importance<br />

of managing manual activities (Work Items)<br />

separately from transactional activities (Invoked<br />

Applications). For example, ensuring the integrity<br />

of executed processes requires both the ability to<br />

initiate compensating actions if a transaction fails<br />

and to identify when a specific activity requires<br />

human intervention. This requires a broader<br />

perspective (i.e., process-wide) than is otherwise<br />

provided from individual transactions.<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Defining <strong>BPM</strong><br />

6<br />

For example, the J2EE framework offers an 'ACID'<br />

test for the integrity of discrete transactions<br />

(defined vis-à-vis the four properties of Atomicity,<br />

Consistency, Isolation, and Durability). This is an<br />

important enabler for the <strong>BPM</strong> system to ensure<br />

the integrity and completion of individual process<br />

steps. However, ACID as a specification does not<br />

adequately address the need for state management<br />

across long-running transactions and business<br />

processes. Similarly, while J2EE session beans<br />

enable state to be held or passed from one<br />

transaction to another, they are not capable of<br />

managing higher-level business logic or flow<br />

control across the execution of a process.<br />

Rather, this level of control and coordination<br />

requires an Execution Engine to manage state<br />

changes generated by each activity instance across<br />

an initiated process. This also allows the<br />

Execution Engine to initiate corrective actions<br />

based on process definitions and business rules.<br />

Figure 1 - Conceptual Model of <strong>BPM</strong><br />

User Environment<br />

Work Items<br />

& Work List<br />

Process Designer<br />

Defines<br />

Process Definition<br />

<strong>BPM</strong> System<br />

Generates<br />

Instantiates<br />

Invokes<br />

Execution Engine<br />

Updates<br />

Activity Monitor<br />

Accesses<br />

This may involve time limits defined within the<br />

process definition to trigger the prompting for<br />

user response, or process logic may require that<br />

routing be changed based on the state changed<br />

in an activity instance.<br />

The ability to manage state and process logic also<br />

allows the Execution Engine to facilitate hand-offs<br />

between process participants (person-to-person)<br />

and manage interactions between users and<br />

applications. For example, when a work item is<br />

delivered to the work list, the Execution Engine<br />

can pass on both the state from the previous<br />

activity and the description of what to do from<br />

the process definition (which may also be<br />

conditionally modified based on the state).<br />

Coordinating hand-offs between users<br />

distinguishes the role of the Execution Engine as a<br />

means of optimizing the performance of processes,<br />

rather than simply executing rules or code. This is<br />

accomplished in part by passing on Englishlanguage<br />

instructions and activity descriptions<br />

within a process object, as well as with manual<br />

Third Party<br />

Applications<br />

& Automated<br />

Activities<br />

Security &<br />

Directory<br />

Services<br />

Validates<br />

Persists<br />

Operational<br />

Data<br />

activities where written instructions<br />

are delivered through a work item.<br />

In this manner, the Execution Engine<br />

allows autonomous execution of<br />

individual activities, while maintaining<br />

the continuity and integrity of the<br />

executed process. While no company<br />

keeps its knowledge workers on a<br />

production line, it should still be<br />

evident how <strong>BPM</strong> enables more<br />

efficient utilization of human resources<br />

by facilitating the execution of work.<br />

Specifically, by bringing work directly<br />

to users with the process context and<br />

task information necessary to execute<br />

each step, rather than having them<br />

waste trying to figure out what to do,<br />

what they need and where to find it.<br />

This also illustrates another key benefit<br />

of <strong>BPM</strong> systems versus the alternative<br />

approaches of application integration<br />

and automation.<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Process Orchestration<br />

7<br />

By facilitating the execution of manual activities,<br />

the <strong>BPM</strong> system is not automating work (as parts<br />

of the process still consist of human work), but<br />

rather it is orchestrating the process.<br />

The Execution Engine manages the sequencing<br />

and distribution of activities, as well as parameters<br />

such as data variables captured or updated, time<br />

constraints on activities, and the state of activities.<br />

Yet the performance of work is left to human<br />

workers, not scripted into application code.<br />

This again presents a subtle yet significant<br />

difference that clearly illustrates the value of <strong>BPM</strong><br />

software. Just as manual activities can be defined<br />

and managed within a process without having to<br />

code every action, processes can be defined within<br />

a <strong>BPM</strong> system based on specific goals and<br />

outcomes without having to explicitly define<br />

every step. This presents the notion of process<br />

orchestration, which differentiates <strong>BPM</strong> from<br />

traditional workflow automation.<br />

Orchestration vs. Automation<br />

The basic notion of orchestration involves the<br />

development of executable processes as a set of<br />

loosely-coupled activities, rather than explicitly<br />

coded steps. Yet the concept of orchestration is<br />

rooted in the fundamental shift of application<br />

design towards the delivery of software as a<br />

service, evolving from the maturation of both J2EE<br />

and Web services. This shift has shaped both the<br />

philosophy and fundamental technologies behind<br />

nearly all software development today.<br />

A simple definition of an application is a set of<br />

activities or use-cases strung together for a<br />

specific purpose or function. The philosophy that<br />

governed application development until only<br />

recently was that of building an amalgamation of<br />

activities through which business and application<br />

logic are inseparable. As is vivid in virtually every<br />

enterprise today, the inevitable result is a rigid and<br />

virtually unmanageable morass of code.<br />

Certainly a better approach is to build as a<br />

collection atomic activities, connected and<br />

coordinated by a separate layer of business logic.<br />

This approach has slowly emerged through the<br />

evolution of N-tier application architectures,<br />

driven by the ability to leverage existing services<br />

from the platform and infrastructure. A necessary<br />

piece missing from component-based platforms<br />

such as J2EE and Microsoft (COM and .NET),<br />

however, is the independent process layer.<br />

Web Services and Process Orchestration<br />

Although componentization and its inherent<br />

benefits of software reuse has been common<br />

within the world of J2EE programming, the<br />

emergence of open standards for integration<br />

(most notably the Web services stack of SOAP,<br />

WSDL, XML and UDDI) presents a standard,<br />

application-independent means for components<br />

to be exposed as a library of invokable services.<br />

Connecting two points (i.e., reading into a single<br />

application from another) is not by itself<br />

revolutionary. What is, however, is building<br />

libraries of components into a vast network of<br />

distributed services. This is the concept of<br />

software virtualization, or what is increasingly<br />

referred to as a Service-Oriented Architecture<br />

(SOA).<br />

While most of the attention focused on SOA<br />

has been largely dominated by Web services,<br />

it is important to realize that these standards<br />

are merely one of the means to an end and not<br />

the end game itself. The true value of SOA comes<br />

from the ability to orchestrate these components<br />

across executable business processes.<br />

Bridging the Islands of Automation<br />

By definition, the use of Web services presumes<br />

that automation is already in place (i.e., Web<br />

services enables connectivity between automated<br />

activities), yet very often this exists as the<br />

proverbial "islands of automation" with no<br />

visibility between them. One system might handle<br />

the shipping documents, with another<br />

coordinating shipping schedules, and another<br />

tracking shipping status. Thus each system is<br />

responsible for a discrete function, yet lacks the<br />

ability to determine if or where a process may have<br />

broken down within any of the related but<br />

otherwise disconnected systems.<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Phases of Orchestration<br />

8<br />

Whether or not your business involves shipping<br />

products, this is nonetheless likely to be a familiar<br />

scenario for any firm that has invested in<br />

automation without orchestration. One approach<br />

to this problem would be to integrate these<br />

otherwise disparate systems with EAI. Yet this<br />

would not address the need for process<br />

coordination, would not help with handling<br />

exceptions, and would still come at a very high<br />

deployment cost. Exposing these application<br />

capabilities as Web services would offer visibility<br />

without the inherent rigidity of EAI, and thus over<br />

the long-term would be more cost effective to<br />

maintain. But Web services alone would do little to<br />

help coordinate processes across these systems.<br />

The First Phase of Orchestration<br />

Most <strong>BPM</strong> systems on the market today can<br />

“consume” Web services. In other words, when<br />

Web services are registered within a standard<br />

directory (UDDI) and wrapped with a standard set<br />

of descriptors (WSDL and XML) they can be<br />

invoked by a <strong>BPM</strong> system as an automated activity.<br />

Many <strong>BPM</strong> systems allow Web services to be<br />

“discovered” from within the process designer and<br />

added to the design palette, such that they can be<br />

added to the process definition without requiring<br />

any other formal integration effort.<br />

The standard Web services interfaces describe to<br />

the <strong>BPM</strong> system how to invoke the service and<br />

what data variables can be sent to and/or from the<br />

service. Depending on the complexity of the<br />

service, these data variables are either mapped<br />

within the process designer or automatically<br />

added to the process definition requiring only<br />

basic validation during the design process.<br />

This represents the significant first stage of<br />

moving from automation to orchestration, by<br />

combining process management with Web<br />

services. It allows for external applications and<br />

data to integrated into executable business<br />

processes with minimal programming effort<br />

compared to what would otherwise be required.<br />

It also leverages the inherent flexibility of Web<br />

services over other integration approaches. Since<br />

it is expressed as a Web service, the service<br />

component is self-describing and self-contained.<br />

Changes within the application (such as adding a<br />

new data table or moving up to a new system<br />

version) will be resolved through the Web services<br />

definition and not require a redesign of the<br />

process. Anyone who has dealt with other forms of<br />

EAI knows full well the problems involved with<br />

application changes breaking integration<br />

connections. In the context of live processes, these<br />

changes will bring work to a screeching halt.<br />

However the combination of Web services and<br />

process management also allows the ability to<br />

design processes based on specific goals and<br />

outcomes, without having to explicitly define every<br />

possible permutation or flow path. Just as with a<br />

manual activity, within an orchestrated process a<br />

step can be defined as "Check Shipment" or "Notify<br />

Receiver" (using the example earlier in this<br />

section) and as such invoke a Web service to<br />

retrieve shipping status or send out a notification<br />

based on a specific set of input variables. To do<br />

this within a traditionally automated process<br />

would require volumes of coding.<br />

The Next Phase of Orchestration<br />

Beyond consuming Web services, the next stage of<br />

orchestration is to use the <strong>BPM</strong> systems<br />

specifically to enable software virtualization.<br />

Although the Web services architecture is<br />

increasingly adopted by application developers<br />

such as SAP and PeopleSoft, as well as platform<br />

vendors such as IBM and Microsoft, the most likely<br />

scenario is that application capabilities desired are<br />

not exposed services and may not have the<br />

descriptive metadata required to easily access<br />

without integration.<br />

This presents two obvious options: using<br />

proprietary application adapters, or building<br />

proprietary integration points. Pursuing either<br />

option outside of a <strong>BPM</strong> system (i.e., for the<br />

purpose of accessing these within process<br />

definitions) is an expensive proposition in terms<br />

of either the cost of application adaptors or the<br />

cost of development time or both.<br />

Addressing this from within a <strong>BPM</strong> system offers<br />

two similar options, yet in most cases at a<br />

much lower price point. This also likely includes<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Orchestration & Virtualization<br />

9<br />

leverage of standard interfaces such as ODBC,<br />

COM/DCOM or J2EE interfaces such as JAAS, JTS,<br />

JDBC, and JNDI. But for proprietary systems other<br />

means of entry are required.<br />

Many <strong>BPM</strong> systems on the market today offer the<br />

option of a pre-integrated library of application<br />

adapters. These carry with them two important<br />

caveats:<br />

1) there is no such thing as a “universal<br />

adapter” as every application requires some<br />

modification of data structures; and<br />

2) adapters are typically designed for a specific<br />

version, which must be synchronized with the<br />

connected applications.<br />

What pre-integrated adapters do allow, however, is<br />

the ability to define integration points within the<br />

process designer and enable easier access to<br />

applications than would be available with<br />

proprietary EAI. Once integrated, this allows<br />

application services to be registered once and<br />

reused throughout process definitions.<br />

The second approach is to enable integration<br />

syntax directly within the process designer. <strong>BPM</strong><br />

systems which offer this capability typically<br />

delineate separate levels of abstraction --- one<br />

which defines business and flow logic, and another<br />

which defines integration logic. These also allow<br />

the registry of application services as individual<br />

components, enabling them to be included within<br />

the designed palette and reused throughout the<br />

process definition.<br />

Certainly, neither approach offers<br />

a “magic bullet” to application integration.<br />

Any integration effort<br />

will require significant development time<br />

and effort to introduce external<br />

application services to process<br />

management. What these<br />

alternatives do offer over other<br />

approaches, however, is specifically<br />

that -- allowing application services<br />

to be introduced to process<br />

management rather than trying to<br />

manage processes around<br />

application functionality.<br />

Figure 2 - An Automated Process<br />

Figure 2 above illustrates a typical<br />

automated process, where each step is<br />

defined explicitly in terms of all possible<br />

permutations and outcomes.<br />

Figure 3 below illustrates the same<br />

process based on an orchestration<br />

strategy, where steps are defined as<br />

autonomous activities or invoked<br />

services. The same level of control flow<br />

is enabled in both examples, however,<br />

the orchestrated process offers a far<br />

more streamlined design and fewer<br />

points of failure (by defining steps in<br />

terms of outcome rather than scripted<br />

rules).<br />

Figure 3 - An Orchestrated Process<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Orchestration Summary<br />

10<br />

Process Orchestration Myths and Realities<br />

The role of orchestration in process management<br />

offers a vivid illustration of evolutionary advances<br />

of current <strong>BPM</strong> software over earlier approaches to<br />

process automation. It should be noted that<br />

orchestration is not a “quick fix” for managing<br />

processes without proper analysis of business<br />

logic.<br />

What an orchestration-based approach does offer<br />

is the ability to manage processes of greater<br />

complexity with far more efficiency than is<br />

otherwise possible with alternative approaches.<br />

The key to this is a modular approach to managing<br />

business rules, relationships, and activities.<br />

Within an automation paradigm, processes are<br />

defined “end-to-end” with all possible paths (or<br />

more commonly a single path) pre-determined.<br />

Thus ‘Step #5’ always follows ‘Step #4’ and precedes<br />

‘Step #6’ even if different instances of an otherwise<br />

standard process may require a different sequence.<br />

Orchestration allows for the sequencing of steps to<br />

be determined during the “run-time” instance of a<br />

process, with paths determined by evolving context<br />

resulting from each new step. Thus the potential<br />

number of paths and outcomes may otherwise be<br />

too complex to define in terms of pre-determined<br />

“If-Then-Else” rules, but may be easily resolved<br />

through human interaction and decision-making.<br />

This highlights the key architectural difference<br />

between automation and orchestration. Given the<br />

inherent complexity and constant changes within<br />

any “real world” business environment, effectively<br />

managing processes requires the agility to shift<br />

with changes in context, rather than always being<br />

bound to the same scripted flow. This requires the<br />

unique ability to define processes as a set of<br />

atomic, goal-based activities with the enforcement<br />

of basic parameters (e.g., time limits, data<br />

variables), while separating the execution logic<br />

activities from the higher-level process definition.<br />

As discussed throughout this section, process<br />

orchestration presents the intersection of process<br />

management and Web services. To be clear,<br />

however, process orchestration should not be<br />

confused with other emerging approaches of Web<br />

services orchestration and choreography. Process<br />

orchestration is not limited to invoking software,<br />

but rather represents a shift from task-based to<br />

goal-oriented process definition. Web services<br />

and other forms of software automation are<br />

utilized through process orchestration, yet not to<br />

the exclusion of manual, human-driven activities.<br />

When examining the process management<br />

software space, differentiation should be made<br />

between automation and orchestration, as well as<br />

between service-oriented processes versus contextdriven<br />

process management focused on<br />

coordinating both manual and automated<br />

activities. <strong>BPM</strong> software products overlap one<br />

or more of the categories defined below.<br />

Process Management<br />

Software Landscape<br />

Orchestration-Centric<br />

Service-Oriented<br />

Supports processes<br />

involving the coordination<br />

of external application<br />

services. Enables both<br />

consuming and generation/<br />

publishing of services from<br />

external applications.<br />

Provides fine grain control<br />

over invoked services.<br />

Workflow-Driven<br />

Supports processes with<br />

a discrete number of<br />

predetermined steps,<br />

paths, and outcomes.<br />

Supports invocation of<br />

external services through<br />

data exchange only.<br />

Context-Driven<br />

Supports process flow paths<br />

dependent on the outcome<br />

of previous steps. Enables<br />

orchestration processes<br />

where all possible path<br />

permutations and process<br />

outcomes cannot be<br />

pre-defined.<br />

Rule-Driven<br />

Supports the ability to<br />

define and resolve<br />

discrete business rules.<br />

Rules engines use<br />

If-Then-Else definitions<br />

to resolve process<br />

conditions and events.<br />

Automation-Centric<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Components of <strong>BPM</strong><br />

11<br />

Whole-Product Requirements:<br />

Market Expectations<br />

In an effort to assess the market’s priorities for<br />

whole-product requirements, respondents were<br />

asked to identify which components they believe<br />

should be either self-contained (i.e., proprietary),<br />

pre-integrated (i.e., third-party component<br />

integrated by vendor) or independent (i.e.,<br />

integration left to deploying organization).<br />

As stated earlier, the combination of these<br />

components represents the basic configuration<br />

of any viable <strong>BPM</strong> solution (with the exception of<br />

the App server). The objective with this research,<br />

however, is to determine the market’s expectation<br />

for best-of-breed software options for <strong>BPM</strong>. Not<br />

surprisingly, two-thirds of respondents view<br />

<strong>BPM</strong> as necessarily independent of the app server<br />

platform (counter to positioning by vendors<br />

such as IBM, Oracle, and BEA Systems).<br />

What is notable among these responses is the<br />

high ranking of business rules engines as a<br />

core component. This runs counter to the<br />

fact that rules engines have been largely<br />

abstracted from application software “stacks”<br />

and are today widely available as independent<br />

services. The likely explanation for this is<br />

the market’s expectations for the ability<br />

to define business rules within<br />

process definitions, using the<br />

process designer.<br />

Many vendors leverage separate rules engines,<br />

rather than seeking to replace already deployed or<br />

otherwise off-the-shelf options available on the<br />

market today.<br />

Combining functions of business rules and<br />

process execution, as well as activity monitoring,<br />

presents the capabilities found in the sub-segment<br />

of “<strong>BPM</strong> engines” on the market today. These<br />

represent an emerging class of software<br />

applications focused specifically on introducing<br />

process execution capabilities to existing<br />

applications and new development initiatives.<br />

Typically independent <strong>BPM</strong> engines work across<br />

multiple deployment environments. This also<br />

means that no deployment environment (i.e., app<br />

server) is provided, and thus anticipates that target<br />

applications will have the necessary deployment<br />

infrastructure in place. Similarly, in contrast with<br />

other offerings that base differentiation on the<br />

completeness of built capabilities, independent<br />

<strong>BPM</strong> engines focus foremost on lightweight<br />

deployment overhead and are designed to leverage<br />

existing infrastructure wherever possible. Through<br />

the use of component-based architectures, many<br />

<strong>BPM</strong> solutions also offer execution engines alone<br />

as OEM offerings for other software vendors.<br />

Necessary Components of<br />

Any <strong>BPM</strong> Solution (ranking by respondents)<br />

Should Be a<br />

Core Component<br />

of Any <strong>BPM</strong> Solution<br />

Vendor Should<br />

Pre-Integrate<br />

Best-of-Breed Alternatives<br />

Definitely Should<br />

Be Indepdent<br />

No Answer<br />

Given<br />

Business Rules Engine 67% 26% 7%<br />

Execution Engine<br />

(manage process instances)<br />

Process Activity Monitor<br />

Directory Services (Define and Manage<br />

Users, Roles, Teams, etc)<br />

Analytical Reporting<br />

& Performance Metrics<br />

Integration Platform<br />

Work Portal<br />

(access <strong>BPM</strong> work list)<br />

Application Server<br />

64% 27% 7% 3%<br />

60% 33% 7% 0%<br />

49% 39% 12% 0%<br />

45% 41% 12% 1%<br />

45% 41% 12% 2%<br />

44% 42% 13% 2%<br />

31% 46% 22% 1%<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


What’s Missing From Current Offerings<br />

12<br />

Market’s Perspective:<br />

As shown at the bottom-right chart, respondents<br />

were asked to indicate from a provided list of<br />

capabilities which of those were perceived to be<br />

missing from current <strong>BPM</strong> solutions. This was a<br />

“check all that apply” choice and as such the most<br />

relevant take-away from this is not individual<br />

numbers but the contrast between each category.<br />

The fact that process model standards featured so<br />

prominently on this (cited by nearly half of all<br />

survey respondents) indicates this to be a<br />

significant issue among <strong>BPM</strong> users and evaluators.<br />

The next two highest ranking responses are related<br />

to process modeling if not interchange standards<br />

directly. Combined these reinforce the importance<br />

of the development environment as well as the use<br />

of <strong>BPM</strong> as tool for enabling process improvement.<br />

This is consistent with the results from a related<br />

question (illustrated below) where respondents<br />

were asked to identify their single greatest priority<br />

for investing <strong>BPM</strong>.<br />

Enterprise-wide Redesign<br />

of Business Processes<br />

Data-Level<br />

Application Integration<br />

Rule-based or Event-Drivern<br />

Document Routing<br />

Enforce Business Rules<br />

With Trading Partners<br />

Define & Map Business<br />

Processes<br />

Monitor Staff Performance<br />

& Activities<br />

Ensure Integrity of<br />

Internet Transactions<br />

Extend Existing Apps<br />

11.0%<br />

18.9%<br />

9.9%<br />

13.9%<br />

9.3%<br />

7.4%<br />

8.8%<br />

8.2%<br />

6.5%<br />

4.1%<br />

4.5%<br />

5.7%<br />

4.0%<br />

4.1%<br />

46.0%<br />

37.7%<br />

2003 Survey<br />

2001 Survey<br />

0% 10% 20% 30% 40% 50%<br />

Respondents were asked to identify<br />

their Single Greatest Priority for <strong>BPM</strong><br />

deployments from the list shown above.<br />

Enterprise-wide process redesign was<br />

favored nearly 5-to-1 over integration or<br />

other choices provided, representing a<br />

significant shift over responses in 2001.<br />

Reality:<br />

Enabling the interchange of process models has<br />

been a priority for workflow and (later) <strong>BPM</strong><br />

vendors since the beginning of this industry<br />

segment. This was one of the initial goals of the<br />

Workflow Management Coalition (WfMC) with a<br />

specification known as “Interface 1” and later the<br />

Workflow Process Definition Language (<strong>WP</strong>DL).<br />

Despite some proof-of-concept demonstrations,<br />

neither standard has featured prominently in <strong>BPM</strong><br />

deployments or offerings.<br />

What has shifted the landscape for interchangeable<br />

process models, however, is the use of XML to<br />

represent process definitions. This has enabled a<br />

number of proprietary interfaces between process<br />

modeling tools and process execution engines, as<br />

well as XML-based standards such as the WfMC’s<br />

XML Process Definition Language (XPDL).<br />

The second wave of process standards was<br />

initiated first by <strong>BPM</strong>I.org with the development of<br />

Business Process Modeling Notation (<strong>BPM</strong>N) for<br />

model definition and Business Process<br />

Management Language (<strong>BPM</strong>L) as declarative<br />

language for process execution. These are<br />

discussed in detail within the section on standards<br />

at the end of this report.<br />

Features Missing From <strong>BPM</strong> Solutions<br />

(percent cited from all respondents)<br />

Standardized Interchange<br />

of Process Models<br />

Business Performance<br />

Management<br />

Richer Graphical<br />

Monitoring & Simulation<br />

Orchestration & Negotiation<br />

Quality of Service<br />

Metrics & Monitoring<br />

Analytical Reporting<br />

& Monitoring Capabilities<br />

Transaction Verification<br />

& Delivery Guarantees<br />

Usage Metering & Billing<br />

34.0%<br />

33.0%<br />

28.0%<br />

27.0%<br />

26.0%<br />

22.0%<br />

18.0%<br />

44.0%<br />

©2003 Delphi Group<br />

0% 10% 20% 30% 40% 50%<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Product Selection Criteria<br />

13<br />

Product Selection Criteria:<br />

How Market Evaluates <strong>BPM</strong> Solutions<br />

Similar to the exercise to identify market<br />

perceptions, respondents were asked to rank the<br />

level of priority for the issues and functions shown<br />

below with regards to product selection criteria.<br />

Of the fifteen points of consideration, responses<br />

were categorized in terms of integration-specific<br />

criteria (see page 16) and overall functionality<br />

(below).<br />

Among issues of overall functionality, those with<br />

highest negative responses (i.e., instances where<br />

the lowest priority is placed) relate to platform<br />

choices, which is understandable given the<br />

inherent polarity of these choices and the fact that<br />

a certain percentage of the population will always<br />

pick one platform over another.<br />

The area of criteria associated with the greatest<br />

priority, however, should not be surprising to<br />

anyone who has gone through a product selection<br />

process, yet may otherwise be dismissed by some<br />

as trivial. The criterion in question is the role of<br />

process modeling environments, which is often<br />

dismissed by industry pundits as immaterial,<br />

compared with issues of server performance and<br />

other more technically oriented matters. Despite<br />

this, Delphi’s field experience has consistently<br />

concurred with the findings of this piece of<br />

research, showing that <strong>BPM</strong> software evaluators<br />

very often place a top priority on process<br />

modeling environments, a factor often<br />

underestimated by software suppliers.<br />

Aligned with process modeling priority is the<br />

need to leverage existing resources and skill sets<br />

(ranked nearly as high). This ties into the fact<br />

that most <strong>BPM</strong> deployments involve the<br />

development of an internal process team and<br />

competency. These teams are typically not lead<br />

by IT (a fact reinforced by findings shown later in<br />

this report), but rather are made up largely of<br />

business process owners and users. For this<br />

reason, process designers must be able to<br />

accommodate non-technical users.<br />

The third highest priority is access to integrated<br />

analytical capabilities and business performance<br />

monitoring, which also ranked as a close third<br />

within whole-product requirements. Again, this<br />

reinforces the notion of <strong>BPM</strong> as a means of<br />

enabling continuous process improvement.<br />

Development trends among <strong>BPM</strong> vendors reflect<br />

this requirement, with an increasing number<br />

including OLAP capabilities and other robust<br />

report functionality within core product offerings.<br />

Overall Functional Priorities Driving<br />

Product Selection Criteria<br />

Graphical Mapping & Monitoring<br />

of Business Process<br />

Leverage of Existing<br />

Resources & Skill Sets<br />

Integrated Analytics Engine, or<br />

Business Performance Monitoring<br />

Support for J2EE App Servers<br />

& Related Standards<br />

Independent Validation of<br />

Project & Software ROI<br />

Acccess to a Published Standard<br />

for Process Modeling & Definition<br />

Support for Microsoft<br />

Infrastructure & Standards<br />

©2003 Delphi Group<br />

Little to No<br />

Importance<br />

Moderately<br />

Important<br />

Very<br />

Important<br />

Single<br />

Greatest<br />

Priority<br />

8% 29% 53% 9%<br />

8% 33% 53% 6%<br />

11% 37% 46% 6%<br />

21% 29% 45% 5%<br />

19% 33% 42% 7%<br />

17% 36% 40% 7%<br />

30% 23% 40% 7%<br />

Direction of Increasing Priority<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Defining Business Logic<br />

14<br />

Business Software for a<br />

Business Audience<br />

As discussed on the previous page,<br />

<strong>BPM</strong> deployments are led<br />

predominantly by line-of-business<br />

managers and business process owners.<br />

This is not to suggest that <strong>BPM</strong> does not<br />

require the involvement of IT and other<br />

technical resources. IT will be required<br />

for <strong>BPM</strong> just as with any other major<br />

software deployment. What is unique<br />

about <strong>BPM</strong>, however, is the much larger<br />

role played by traditional business<br />

(i.e., non- technical) personnel.<br />

The two most significant components of<br />

a <strong>BPM</strong> deployment are defining business logic and<br />

identifying integration points. This requires the<br />

ability to express <strong>BPM</strong> deployment requirements<br />

in non-technical terms, including process flows as<br />

well as complex business rules and relationships.<br />

This requirement highlights the importance of the<br />

process designer and development environment,<br />

which holds the potential to be the greatest point<br />

of failure in any <strong>BPM</strong> deployment.<br />

If processes are not properly articulated and<br />

validated in the design phase, they will of course<br />

fail upon deployment. Treating end users as the<br />

test bed for processes not properly validated will<br />

result in significant culture resistance and<br />

reluctance to cooperate with further iterations.<br />

Senior Management, Not CIO<br />

Individual Process Owners<br />

or Line-of-Business Managers<br />

©2003 Delphi Group<br />

Individual Process Owners<br />

or Line-of-Business Managers<br />

Enterprise Architect<br />

E-Business Leader<br />

Other Business Analyst(s)<br />

Outside Consultants<br />

Other (please specify)<br />

Internal Programmers<br />

©2003 Delphi Group<br />

Project Sponsor for <strong>BPM</strong> Investments<br />

CIO (or equivalent)<br />

Other<br />

Marketing<br />

E-Business Leader<br />

2.6%<br />

2.1%<br />

7.2%<br />

11.6%<br />

30.0%<br />

0% 10% 20% 30% 40% 50%<br />

46.5%<br />

Responsibility for Defining<br />

Business Rules & Process Logic<br />

2%<br />

2%<br />

4%<br />

9%<br />

12%<br />

17%<br />

It is for these reasons that proper definition of<br />

process logic is necessary for the success of any<br />

<strong>BPM</strong> deployment, and why input from business<br />

process owners is so crucial.<br />

As stated earlier, most <strong>BPM</strong> deployments involve<br />

the development of cross-functional (i.e., IT and<br />

business) process teams. The lingua franca for<br />

these teams may initially be flow charting tools,<br />

but at some point (ideally earlier than later) this<br />

needs to switch to the actual process design<br />

environment used to create process definitions.<br />

The later in the deployment process this segue<br />

occurs, the greater the probability for the<br />

misinterpretation of business requirements as<br />

they are translated into process definitions.<br />

55%<br />

This has led some firms to use<br />

modeling methodologies such as UML,<br />

that are designed for enterprise and<br />

application modeling. The caveat with<br />

methodologies such as UML, or even<br />

process-specific methods such as IDEF,<br />

is that these do not allow for fine grain<br />

articulation of product-specific process<br />

definitions, and thus run a similar risk<br />

of misinterpreting business<br />

requirements. A better approach is to<br />

select a <strong>BPM</strong> solution that supports<br />

business-level process modeling within<br />

the same environment as more granular<br />

integration and application syntax.<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


<strong>BPM</strong> Benefits<br />

15<br />

What’s in it for IT?<br />

Respondents were asked to identify<br />

(separately) the anticipated IT-specific<br />

benefits and overall functional benefits<br />

anticipated through deploying <strong>BPM</strong> software.<br />

The clear winner among IT benefits is the<br />

ability to integrate internal applications.<br />

As this admittedly presents only a limited<br />

perspective (respondents were forced to pick<br />

one answer from limited options) further<br />

definition of integration priorities are<br />

provided on the next page.<br />

What these responses do indicate more than<br />

anything else, however, is the education<br />

challenge facing <strong>BPM</strong> vendors. Integration<br />

ranked fairly low among top priorities as<br />

shown earlier in this report. Yet a challenge the<br />

<strong>BPM</strong> market has faced is lack of recognition for<br />

other IT-related benefits than integration. These<br />

include easier (and cheaper) maintenance of<br />

enterprise software, through the use of<br />

orchestration capabilities (versus custom-coding)<br />

to expose application services. As well as the<br />

ability to develop new capabilities without the<br />

comparatively higher expense involved with<br />

building these within existing application<br />

environments.<br />

As many CIO’s budgets are today consumed<br />

largely by application integration and<br />

maintenance costs, <strong>BPM</strong> offers a<br />

high-leverage opportunity for<br />

addressing new problems such as<br />

Sarbanes-Oxley compliance (which<br />

now puts CIOs on the line for<br />

accounting practices). From a<br />

business user perspective, the<br />

greatest benefit of <strong>BPM</strong> is seen as<br />

the ability to manage business rules<br />

and process definitions without<br />

having to rely on IT for every<br />

required change. This too should be<br />

viewed by IT as an inherent benefit,<br />

as well as a means for enabling<br />

orchestration over automation (i.e.,<br />

leveraging existing automation<br />

within applications).<br />

Change Business & Processs Rules<br />

Within Enterprise Software<br />

Without IT Intervention<br />

Manage Exceptions<br />

Within Automated Processes<br />

Ability To Visualize, Simulate,<br />

& Trouble-Shoot<br />

Business Processes<br />

Automate Repetitive Tasks<br />

& Accelerate Process Cycle Times<br />

Single Greatest Functional<br />

Advantage Anticipated from <strong>BPM</strong><br />

Monitor Personnel<br />

& Process Performance<br />

©2003 Delphi Group<br />

©2003 Delphi Group<br />

Single Greatest IT-Specific Benefit<br />

Anticipated from <strong>BPM</strong><br />

Cost-Effective Integration<br />

of Internal Applications<br />

Accelerate Deployment<br />

of New Applications<br />

Reduce Total Cost of Ownership<br />

of Existing Application<br />

Integration of Applications<br />

Outside of The Firewall<br />

Add "State" to<br />

Stateless Applications<br />

Ability to Reuse Software Code<br />

Other<br />

Other<br />

2%<br />

4%<br />

8%<br />

11%<br />

10%<br />

10%<br />

16%<br />

0% 10% 20% 30% 40% 50%<br />

15%<br />

13%<br />

17%<br />

41%<br />

Respondents were asked to pick the<br />

single greatest advantage from two lists<br />

of IT-related and general business<br />

functional advantages. While cost-effective<br />

integration is clearly favored among<br />

IT benefits, business benefits are much<br />

more evenly split.<br />

24%<br />

28%<br />

0% 5% 10% 15% 20% 25% 30%<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Integration Priorities<br />

16<br />

Integration Priorities:<br />

The Market’s Perspective<br />

When asked to prioritize integration-specific<br />

product selection criteria, respondents indicated a<br />

bias favoring cross-application process integration.<br />

Within the language of the survey, the word<br />

“orchestration” was not used explicitly, as this is a<br />

relatively new term which is open to multiple<br />

interpretations. Yet the desire for orchestration as<br />

described and defined in this report is clearly<br />

favored over simply enabling data-level integration<br />

(both in the results shown below and indicated<br />

earlier).<br />

What is equally clear, is that integration in general<br />

remains a significant priority in most <strong>BPM</strong><br />

deployments. As is illustrated in several different<br />

charts and findings, integration only falls when<br />

respondents are asked to identify top drivers of<br />

<strong>BPM</strong> initiatives (i.e., where the choice of<br />

integration is mutually exclusive of other choices).<br />

This illustrates a central message of this report,<br />

that <strong>BPM</strong> is not about enabling point-to-point<br />

connections between applications, but rather the<br />

ability to define and orchestrate business<br />

processes that take advantage of capabilities and<br />

existing automation within multiple applications.<br />

Where respondents appear equally split or<br />

otherwise indifferent within the findings<br />

illustrated below is when it comes to whether or<br />

not <strong>BPM</strong> should be a component of integration<br />

Connecting Otherwise Disparate Processes<br />

Across Multiple Software Applications<br />

software or offered independent of any EAI or<br />

integration platform. This is a subtle distinction<br />

in the context of a high-level assessment of <strong>BPM</strong>.<br />

Yet for practical purposes, <strong>BPM</strong> offered as a<br />

component of EAI will not allow for necessary<br />

flexibility to enable process orchestration.<br />

Architecturally, <strong>BPM</strong> needs to be an independent<br />

layer which takes advantage of integration<br />

capabilities, either internally or in partnership<br />

with third-party EAI offerings. Otherwise, the<br />

ability for <strong>BPM</strong> services to be invoked by both<br />

external applications and user environments<br />

(work lists) will be compromised.<br />

Most <strong>BPM</strong> offerings on the market today present<br />

some level of integration capability, with varying<br />

levels and approaches. For <strong>BPM</strong> initiatives where<br />

external applications play a large role at multiple<br />

points across a process, the ability to define<br />

invocation capabilities within process definitions<br />

will play a large role in the resulting deployment’s<br />

agility and performance. For processes that are<br />

largely collaborative or “people-centric” in design,<br />

the ability to read data from external applications<br />

may be sufficient. For enabling orchestration<br />

across multiple application services, however, the<br />

<strong>BPM</strong> solution used must allow for fine grain<br />

articulation of integration and invocation syntax.<br />

Integration-Specific Priorities<br />

Driving Product Selection Criteria<br />

Little to No<br />

Importance<br />

Moderately<br />

Important<br />

Very<br />

Important<br />

Single<br />

Greatest<br />

Priority<br />

6% 24% 57% 13%<br />

Integrating <strong>BPM</strong> Within<br />

Existing Applications & Initiatives<br />

Ability to Access Outside<br />

Data Within the <strong>BPM</strong> Environment<br />

Direct Access to API<br />

Set of <strong>BPM</strong> Engine<br />

<strong>BPM</strong> as a Component of<br />

or Integrated With EAI<br />

Independent of Any Single Apps Server,<br />

EAI Tool, or Enterprise Application<br />

8% 23% 57% 13%<br />

8% 26% 57% 9%<br />

13% 36% 45% 6%<br />

20% 35% 40% 5%<br />

22% 36% 37% 5%<br />

Transaction Processing<br />

& Load-Balancing<br />

19% 42% 37% 3%<br />

©2003 Delphi Group Direction of Increasing Priority<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Requirements for Software Suppliers<br />

17<br />

Who Gets The Deal?<br />

The market for <strong>BPM</strong> software remains price<br />

sensitive, with the most common answer by<br />

respondents that vendor selection will be made<br />

according to the lowest price. This may be<br />

disheartening to most <strong>BPM</strong> vendors, yet it remains<br />

a reality for most software segments as<br />

IT budgets continue to shrink relative to available<br />

dollars (not otherwise earmarked for maintenance<br />

and integration). This number may be understated<br />

due to best intentions by some respondents,<br />

who may change their tune when ultimately forced<br />

to justify their investments. Thus the percent of<br />

purchases made based on the lowest bid may<br />

actually be larger than is indicated here.<br />

One significant shift seen in the 2003 results over<br />

2001, however, is the favoring of large brand<br />

vendors. This too reflects macro trends across<br />

the software industry, with many smaller vendors<br />

complaining of what felt to be the unfair scrutiny<br />

of their size and viability. Yet for the <strong>BPM</strong> market<br />

it is too soon for a “Big 3” set to dominate.<br />

Innovation traditionally flows from upstarts, not<br />

incumbents. This has been no different for the<br />

<strong>BPM</strong> software industry, where large platform<br />

vendors such as Microsoft, IBM, and Oracle (e.g.,<br />

the traditional software Big 3) have remained<br />

notably deficient in process management<br />

capabilities (in particular relative to the offerings<br />

of much smaller vendors).<br />

The lack of presence by large<br />

platform vendors reflects the<br />

generally nascent state of the <strong>BPM</strong><br />

sector (a market arguably only a few<br />

years old) and the lack of critical<br />

mass among large system buyers<br />

(see pages 19-20 for the size of total<br />

and individual spending). One factor<br />

already shaping the competitive<br />

landscape, however, is the emergence<br />

of standards driven by the larger<br />

platform players.<br />

Most notable is BPEL or Business Process Execution<br />

Language, submitted to the standards body OASIS<br />

by vendors including Microsoft, IBM, BEA Systems,<br />

and SAP. Due to its backing by these large vendors,<br />

as well as its growing support by “pure play” <strong>BPM</strong><br />

vendors, BPEL is expected by many to emerge as<br />

the dominant <strong>BPM</strong> standard. As an adopted<br />

standard, BPEL would allow a single process<br />

definition to run on multiple execution engines, as<br />

well as to exchange process objects between live<br />

processes running on different systems.<br />

If BPEL or another standard does mature to allow<br />

true interchangeability between <strong>BPM</strong> systems, this<br />

will put pressure on all <strong>BPM</strong> players to<br />

differentiate based on valued-added capabilities.<br />

Similar to the role that J2EE played on the until<br />

recently proprietary app server market, a single<br />

<strong>BPM</strong> standard should foster innovation by<br />

removing the threat to software buyers of locking<br />

into proprietary investments. If in this way<br />

competitive pressure (vis-a-vis buying criteria)<br />

shifts from company size to price and<br />

functionality, the results will be a greater number<br />

of affordable options for <strong>BPM</strong> buyers.<br />

Platform vendors have the opportunity to<br />

undercut smaller competitors by subsidizing the<br />

cost of <strong>BPM</strong> software. But opportunities remain for<br />

smaller vendors to differentiate on the basis of<br />

lower overall cost of ownership, with products that<br />

are easier and cheaper to deploy and maintain (see<br />

page 20 for a break out of the total cost of <strong>BPM</strong>).<br />

Financial Criteria for <strong>BPM</strong> Software Suppliers<br />

Will Work With Whoever Provides<br />

the Best Combination Of Price & Service<br />

Will Prioritize Selection Criteria<br />

According to Product<br />

& Organizational Capabilities<br />

Only Vendors With<br />

Dominant Brand Name<br />

Only Vendors With<br />

There is an Existing<br />

Commerial Relationship<br />

Only Publicly-Traded Vendors<br />

2%<br />

3%<br />

12%<br />

12%<br />

17%<br />

24%<br />

23%<br />

32%<br />

2003 Survey<br />

2001 Survey<br />

38%<br />

36%<br />

©2003 Delphi Group<br />

0% 10% 20% 30% 40%<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Current State of <strong>BPM</strong> Deployments<br />

18<br />

The Market Comes of Age --<br />

But a Business Case Still Must be Made<br />

While still in an early stage, the <strong>BPM</strong> market<br />

appears to be close to reaching critical mass, with<br />

the number of early adopter deployments nearly<br />

doubling since the 2001 survey (indicated by 20%<br />

of respondents indicating they are using <strong>BPM</strong><br />

already). The high number of firms in the early<br />

stage of evaluation (nearly half of all respondents)<br />

indicates that the market is on the cusp of a break<br />

out. If buyers move as indicated below, 75% of all<br />

firms responding to this survey will have invested<br />

in <strong>BPM</strong> inside of the next three years.<br />

The single greatest factor suppressing market<br />

expansion today is the lack of clearly defined<br />

business case. Combined with the lack of project<br />

sponsors (a closely related issue) the need to make<br />

the case for <strong>BPM</strong> investments has held back more<br />

than 60% of firms who have not yet invested. This<br />

illustrates the education challenge facing the<br />

supply-side of the <strong>BPM</strong> market. It is of greater<br />

importance today than price or functionality.<br />

<strong>BPM</strong> vendors have the onus of proving the<br />

business case for any investment regardless of<br />

price. An effective vehicle for this is the delivery of<br />

packaged solutions able to demonstrate very quick<br />

results, such as Sarbanes-Oxley or USA PATRIOT<br />

Act compliance or domain-specific solutions such<br />

as those for telecom provisioning . These are<br />

focused apps built on top of a core <strong>BPM</strong> platform,<br />

offering near turn-key deployment now, with a<br />

defined path for later expanding into other areas.<br />

Investment Timing for Firms Without<br />

Current Deployments or Initiatives<br />

40%<br />

34%<br />

Using <strong>BPM</strong><br />

Software Already<br />

State of <strong>BPM</strong> Deployments<br />

Final Stages<br />

of Deployment<br />

Early Stage<br />

of Evaluation<br />

No Plans<br />

for <strong>BPM</strong><br />

No Business<br />

Case<br />

No Project<br />

Sponsor<br />

9%<br />

4%<br />

12%<br />

20%<br />

23%<br />

29%<br />

0% 10% 20% 30% 40% 50% 60%<br />

©2003 Delphi Group<br />

Reasons Cited for<br />

Not Investing in <strong>BPM</strong><br />

28.0%<br />

24.0%<br />

28.0%<br />

2003 Survey<br />

2001 Survey<br />

Respondents were asked to identify their<br />

current state of <strong>BPM</strong> deployments, and<br />

where none were planned or in place,<br />

the single greatest reason for this.<br />

Findings were largely consistent<br />

year-over-year, with the most dramatic<br />

changes relative to firms who have<br />

invested since the 2001 survey, and the<br />

growing need for a business case<br />

among those who haven’t.<br />

48%<br />

55%<br />

36.7%<br />

Will Use Existing<br />

Apps/Platform<br />

17.6%<br />

18.3%<br />

9%<br />

< 6<br />

Months<br />

7-12<br />

Months<br />

2-3<br />

Years<br />

10%<br />

4-5<br />

Years<br />

1% 6%<br />

> 5<br />

Years<br />

Never<br />

Vendors Unable<br />

to Meet Needs<br />

Other<br />

14.9%<br />

9.7%<br />

6.8%<br />

16.1%<br />

2003 Survey<br />

2001 Survey<br />

0% 10% 20% 30% 40% 50%<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Investment Size & Pricing Models<br />

19<br />

How Big a Deal?<br />

Another sign of market maturity is the shift<br />

in anticipated spending reported in the “less than<br />

$100,000” range to anticipated spending between<br />

$1 million and $5 million. An investment of less<br />

than $100,000 over a three year period will<br />

certainly pay for a pilot, but not much more than<br />

a small deployment. So the decline in category is<br />

a positive sign for the market overall.<br />

Investment plans have remained relatively<br />

consistent, with the subtle shifts favoring larger<br />

transaction sizes. This is a forward-moving trend,<br />

and is consistent with what should be seen with<br />

any market maturity cycle. Where changes are<br />

most pronounced, however, is in relation to<br />

anticipated pricing models. The percent of<br />

respondents whose preferred pricing model is<br />

by the CPU more than doubled over 2001 findings.<br />

This trend is consistent with the direction of most<br />

vendors, particularly with architectures based on<br />

J2EE. Yet those citing preference for pricing per<br />

user also increased by more than 60% since 2001.<br />

Total <strong>BPM</strong> Software-Only Spending<br />

Anticipated Over the Next Three Years<br />

Nothing<br />

< $100,000<br />

$101,000 -<br />

250,000<br />

$251,000 -<br />

500,000<br />

$501,000 -<br />

1,000,000<br />

$1,001,000 -<br />

2,500,000<br />

$2,500,000 -<br />

5,000,000<br />

> $5,000,000<br />

15.7%<br />

15.4%<br />

20.9%<br />

19.7%<br />

11.3%<br />

10.3%<br />

9.6%<br />

12.0%<br />

5.2%<br />

2.6%<br />

5.2%<br />

0.9%<br />

2.0%<br />

5.1%<br />

30.1%<br />

34.2%<br />

2003 Survey<br />

2001 Survey<br />

0% 10% 20% 30% 40% 50%<br />

©2003 Delphi Group<br />

These price increases are at the expense of fixedprice<br />

and subscription based pricing (including<br />

SLAs). This trend may reverse or increase, based<br />

in the effect of new accounting regulations<br />

(Sarbanes-Oxley most notably) that alter the<br />

manner in software and other expenses must be<br />

amortized and recognized. At this point in time<br />

it is our expectation that subscription pricing<br />

(e.g., delivery of software as a service) will in fact<br />

increase in popularity as a result of these<br />

regulations.<br />

Single, fixed price<br />

for development environment<br />

Price per CPU<br />

Service Level Agreement<br />

Price per user<br />

Subscription fee based<br />

on processing volume and usage<br />

Price per process managed<br />

Preferred Pricing Model<br />

for <strong>BPM</strong> Investments<br />

10.2%<br />

22.0%<br />

27.1%<br />

21.4%<br />

20.0%<br />

23.7%<br />

19.4%<br />

11.9%<br />

12.5%<br />

2003 Survey<br />

17.8%<br />

2001 Survey<br />

4.6%<br />

3.4%<br />

Trends in transaction size indicate market<br />

growth over 2001 survey findings. Within the<br />

2003 survey, almost 90% of respondents cited<br />

<strong>BPM</strong> investments as either generating profits or<br />

at least netting out with expenses. This indicates<br />

expectations for positive ROI on <strong>BPM</strong>, but is also<br />

consistent with an expectation for deploying<br />

<strong>BPM</strong> as part of a for-profit or cost-cutting<br />

initiative, such as Business Process Outsourcing.<br />

Financial Role for <strong>BPM</strong><br />

Deployments<br />

Cost-Neutral<br />

(42%)<br />

Cost Center<br />

(19%)<br />

Profit Center<br />

(39%)<br />

©2003 Delphi Group<br />

0% 10% 20% 30% 40% 50%<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


<strong>BPM</strong> Market Size & Momentum<br />

20<br />

How Big is the Market?<br />

Respondents indicated anticipated spending for<br />

2003 totaling approximately $550 million, with<br />

slightly less than $150 million expected to be spent<br />

on software alone, with the balance spent on<br />

services and hardware.<br />

Delphi’s estimate for the total 2003 <strong>BPM</strong> market is<br />

$500 million for software only, thus the 500<br />

respondents in this survey would represent slightly<br />

less than a quarter of total spending. This is<br />

conceivable, as the total market will not likely see<br />

more than 2,000 total sales transactions in 2003<br />

(half of that number, or roughly 1,000 total<br />

transactions is more likely for 2003 <strong>BPM</strong> sales).<br />

Anticipated spending follows roughly a 1-to-3 ratio<br />

of dollars spent on software versus hardware and<br />

services. The single largest area of anticipated<br />

spending is integration services. This expectation<br />

is justified when viewed through the lens of<br />

historic deployment costs of <strong>BPM</strong> models that rely<br />

heavily on new EAI infrastructure. For more<br />

contemporary approaches which leverage<br />

capabilities such as API introspection, the<br />

integration cost should be much lower, closer to<br />

a 1-to-1 software/services ratio.<br />

The average project size for 2003 is roughly<br />

$300,000 with a range of less than $100,000 and<br />

a handful of projects expected to be greater than<br />

$5 million. A few dozen projects were indicated in<br />

the $2 - $5 million range, which is consistent with<br />

the activity Delphi has seen reported by vendors<br />

in this space. Overall, we anticipate that projects<br />

in the $250,000 - $500,000 range will likely be<br />

initial deployments with the potential to grow<br />

beyond $1 million in the next 2-3 years.<br />

While it is possible to identify a mathematical<br />

average, it is not as easy to identify a “typical”<br />

project size or to set expectations for prospective<br />

buyers as to what they should anticipate spending.<br />

This number will vary according to the type of<br />

deployment, the up-front design and analysis, and<br />

of course the size of the deployment in terms of<br />

active users. But it can be assumed that anything<br />

more than a small pilot will cost $100,000 or more<br />

in software and at least as much in services.<br />

Total Anticipated 2003 Spending<br />

on <strong>BPM</strong> = $550 Million<br />

Integration<br />

Services<br />

(39%)<br />

Other<br />

Consulting<br />

(13%)<br />

Software<br />

(26%)<br />

Hardware<br />

(21%)<br />

Respondents were asked to identify the<br />

total dollars anticipated to be spent on:<br />

<strong>BPM</strong> Sofware (not including EAI, app<br />

servers and other infrastructure);<br />

Integration Services; Other Consulting<br />

Services; and new Hardware specifically<br />

required for anticipated or current <strong>BPM</strong><br />

deployments. Answers were aggregated<br />

to provide the break out illustrated above,<br />

and individually analyzed to provide the<br />

ratios shown below.<br />

Services/Hardware<br />

vs. Software<br />

Average<br />

Multiple<br />

Max<br />

Multiple<br />

Standard<br />

Deviation<br />

Software/<br />

Hardware<br />

1.32<br />

10.00<br />

1.38<br />

Software/<br />

Services<br />

2.42<br />

20.00<br />

2.89<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Platform Expectations<br />

21<br />

Deployment Platforms for Current<br />

and Future <strong>BPM</strong> Initiatives<br />

Respondents were asked a series of questions<br />

relative to their preferences and expectations for<br />

<strong>BPM</strong> deployment platforms. Of these (but not<br />

charted here) was whether or not <strong>BPM</strong> will remain<br />

independent of operating system or app server<br />

platforms. Approximately two-thirds indicated<br />

<strong>BPM</strong> will be absorbed by the platform within 3<br />

years, with the same percentage (although a<br />

different set of respondents) answering that such<br />

should happen based on their own preferences.<br />

Until this convergence occurs, which although an<br />

inevitability will not likely happen for at least 2-3<br />

years, the deployment environment for <strong>BPM</strong><br />

solution remains a consideration. With the 2003<br />

survey we saw a subtle drifting in the direction of<br />

J2EE. Yet the greatest shift is seen in the direction<br />

of heterogenous environments. Nearly 60% of<br />

respondents indicated different front-end<br />

environments from their preferred or anticipated<br />

back-end platforms.<br />

The increased reliance of XML for storing process<br />

definitions, the complete abstraction of the data<br />

management layer, and the growing leverage of<br />

Web services is starting to obviate the platform<br />

limitations which dominated just a few years. At<br />

this stage we firmly maintain that this is no single<br />

“right” platform choice, and that the pursuit of<br />

heterogeneous environments are a realistic goal.<br />

We have seen successful deployments co-mingle<br />

Windows infrastructure with J2EE applications.<br />

Examples include the integration of<br />

Windows-based SAP deployments with J2EE-based<br />

<strong>BPM</strong> servers, as well as the use of UNIX servers for<br />

transactional data and Windows or Web-based<br />

front-end environments. We have also seen cases<br />

where a premature “rush to J2EE” was apparent,<br />

and unnecessary in light of Windows-based<br />

options that also offer a reliable migration path to<br />

J2EE. This opportunity should be further<br />

enhanced by the maturing of .NET, which has<br />

recently started to affect <strong>BPM</strong> solution choices (by<br />

offering components within a Microsoft<br />

environment comparable to those seen with J2EE).<br />

J2EE/Web-only<br />

Web front-end,<br />

Microsoft back-end<br />

Microsoft front-end,<br />

J2EE back-end<br />

Web front-end,<br />

UNIX back-end<br />

Microsoft Only<br />

Anticipated Platform for<br />

Future <strong>BPM</strong> Deployments<br />

16.8%<br />

16.8%<br />

15.0%<br />

26.9%<br />

24.5%<br />

0% 10% 20% 30% 40% 50%<br />

Respondents were asked to identify the<br />

anticipated platform for future <strong>BPM</strong><br />

deployments. Where an app server(s) is<br />

expected to be used, respondents were<br />

asked to indicate which from the list<br />

presented below. What is most notable<br />

about the 2003 findings was the large<br />

percentage using open source options of<br />

JBoss or Oracle 9iAS (who licenses the<br />

Orion open source app server).<br />

IBM WebSphere<br />

BEA WebLogic<br />

Oracle 9iAS*<br />

JBoss*<br />

Sun ONE/iPlanet<br />

Other<br />

*Not Identified in the 2001 Survey<br />

App Server Choices for<br />

Future <strong>BPM</strong> Deployments<br />

3.4%<br />

7.3%<br />

17.0%<br />

14.1%<br />

23.8%<br />

30.3%<br />

22.9%<br />

23.9%<br />

34.5%<br />

39.4%<br />

2003 Survey<br />

2001 Survey<br />

0% 10% 20% 30% 40% 50%<br />

©2003 Delphi Group<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


<strong>BPM</strong> Standards<br />

22<br />

The Role of Standards in <strong>BPM</strong><br />

If Samuel Clemens (a/k/a Mark Twain) were alive<br />

today, he might observe that "The nice thing about<br />

standards is that there are so many to chose from."<br />

The notion of a standard in software development<br />

implies as single method or specification, yet with<br />

every new market multiple "standards" always<br />

emerge. Over time these rearrange like soap<br />

bubbles on water, with overlapping specs merging<br />

into a single standard, while some specialized<br />

standards remain intact on their own.<br />

As firms first seek to push the market on their way,<br />

most new software segments initially see several<br />

redundant standards emerge. Yet the only real<br />

value in a standard is when the market picks one<br />

common approach, a process typically hastened by<br />

the support of "gorilla" vendors backing one<br />

standard and forcing competitors to follow suit.<br />

The <strong>BPM</strong> sector has seen a number of standards<br />

and specifications in its space. These include those<br />

from the Workflow Management Coalition, such as<br />

XPDL, Wf-XML and SWAP, to those designed for<br />

Web Services choreography (notably WSCI and<br />

BPEL), as well as the core Web services standards<br />

of SOAP, WSDL, and UDDI. Add to this a mix of<br />

public interface standards such as ICE,<br />

CommerceNet, RosettaNet, and OASIS’ ebXML BPSS<br />

(Business Process Specification Schema).<br />

It is this deluge of standards that many firms in a<br />

recent Delphi survey cited as their single greatest<br />

criticism of the software standards process, such<br />

that buyers (the most vote in a standards process)<br />

have not yet fully weighed-in on the <strong>BPM</strong><br />

standards front. Yet another problem, as the<br />

findings below clearly illustrate, is that the<br />

market's awareness of <strong>BPM</strong> standards remains<br />

fairly low. Respondents were asked to rank their<br />

perceived value of the list of standards and<br />

consortia shown in the chart to the right.<br />

We admittedly mixed apples and oranges in this<br />

list, as Workflow Management Coalition (WfMC) is<br />

not a standard, but is responsible for a number of<br />

standards/specifications such as XPDL. The same<br />

is true for <strong>BPM</strong>I.org and <strong>BPM</strong>L. UML 2.0 is not<br />

specific to <strong>BPM</strong>, but often shows in requirements<br />

documents for <strong>BPM</strong> (albeit not always advisedly).<br />

Yet rather than anointing any single standard, the<br />

goal of this exercise was to determine the market's<br />

attitudes towards the discrete set of standards and<br />

consortia most commonly associated with <strong>BPM</strong>.<br />

Findings were generally consistent with our own<br />

expectations. Standards in place the longest have<br />

the greatest perceived benefit, yet in general<br />

consortia such as the WfMC and BMPI.org still<br />

have a significant education burden in front of<br />

them before they are viewed as "must haves"<br />

among buyers. One surprise, however, was that<br />

for an audience so overwhelmingly predisposed<br />

to process management, both WfMC and BPEL<br />

ranked lower than we would have expected.<br />

In fairness to the WfMC, they received the highest<br />

vote of confidence from respondents. In most<br />

competitive bids Delphi Group has seen in the last<br />

12-18 months, membership in the WfMC is a very<br />

common requirement for <strong>BPM</strong> software suppliers.<br />

As such it was surprising to see the fine degree of<br />

differentiation among respondents’ rankings<br />

between this decade old organization and<br />

<strong>BPM</strong>I.org, who has been in place less than 3 years.<br />

It would be premature to announce the WfMC’s<br />

decline. Yet the close proximately of its rankings to<br />

<strong>BPM</strong>I undoubtedly will offer fodder to those<br />

pushing the notion of the younger consortia as the<br />

WfMC successor. This argument is largely rooted<br />

Workflow<br />

Management<br />

Coalition<br />

UML 2.0<br />

<strong>BPM</strong>I.org<br />

<strong>BPM</strong>L<br />

RosettaNet<br />

BPEL<br />

XPDL<br />

©2003 Delphi Group<br />

Recognition of Existing <strong>BPM</strong><br />

Standard Initiatives<br />

Greatest<br />

Importance<br />

Moderate<br />

Importance<br />

15% 58% 27%<br />

14% 60% 25%<br />

12% 54% 34%<br />

11% 49% 41%<br />

11% 64% 25%<br />

8% 43% 48%<br />

7% 40% 53%<br />

Little to<br />

No Importance<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


<strong>BPM</strong> Standards<br />

23<br />

in the notion that <strong>BPM</strong>I is more contemporary in<br />

its focus and approach, while the WfMC is<br />

concerned more with earlier workflow<br />

management models. There is some merit to that<br />

argument, yet there is significant parallelism<br />

between their efforts.<br />

For most of its existence, the WfMC’s focus has<br />

been the development of five interfaces between<br />

application connection points, similar to what is<br />

illustrated in Figure 1 on page 6. These have<br />

evolved to incorporate the use of XML and other<br />

standards that have evolved over the last decade.<br />

The goal of <strong>BPM</strong>I has been largely the same --<br />

to develop three specific standards (<strong>BPM</strong>L, <strong>BPM</strong>N,<br />

and BPQL) which allow a common set of definition<br />

to interoperate across any compliant <strong>BPM</strong> system.<br />

In essence, to define the components of a<br />

“standard” <strong>BPM</strong>S based on the three constituent<br />

standards. Although the two sets of specifications<br />

are in many ways competitive, the two groups have<br />

collaborated on making them interoperable.<br />

Although it is certainly possible that both will<br />

continue to co-exist, a more likely scenario is that<br />

key specifications are extracted from the larger<br />

“stack”, such as WfMC’s XPDL co-existing with<br />

<strong>BPM</strong>I’s <strong>BPM</strong>L (which is already under discussion<br />

by the two groups). Also shaping this is whether<br />

BPEL emerges as the winning standard for process<br />

execution, which seems likely will happen.<br />

Pronounced “bipple” it is officially named<br />

BPEL4WS or “business process execution language<br />

for Web services” but more popularly known as<br />

BPEL. BPEL v1.0 was initially submitted for public<br />

review by Microsoft, IBM and BEA Systems in<br />

August 2002, replacing the combined work of<br />

Microsoft's XLANG and IBM's WSFL. Following<br />

some concern within the market over large<br />

vendors dominating the specification, a coalition<br />

of vendors including IBM, Microsoft, BEA Systems,<br />

and SAP submitted BPEL v1.1 to the standards<br />

body OASIS in April of 2003.<br />

BPEL is declarative language for describing<br />

process definitions, meaning that it allows<br />

compliant process execution engines to run BPEL<br />

definitions as executable code. In this way it is<br />

similar in its design and intent to <strong>BPM</strong>L, yet BPEL<br />

lacks some of former’s finer grain process syntax.<br />

This is an advantage for <strong>BPM</strong>L today, however, it is<br />

by no means one that cannot be resolved over time<br />

as the specification evolves.<br />

In the case of BPEL and <strong>BPM</strong>L, both lack the<br />

complexity of syntax to describe the full set of<br />

capabilities for every <strong>BPM</strong> solution. Both would<br />

need to further evolve to see market acceptance as<br />

a true standard among <strong>BPM</strong> vendors.<br />

This fact is evident in the initial coalition of<br />

vendors behind BPEL, who are application and<br />

infrastructure vendors (SAP, Microsft, BEA Systems,<br />

et al) and as such cannot design or support the<br />

more complex processes developed by <strong>BPM</strong><br />

vendors. Since its acceptance by OASIS, however,<br />

<strong>BPM</strong> vendors already aligned with <strong>BPM</strong>I.org in<br />

what is more likely a legitimate move to align the<br />

two, rather than a matter of defection.<br />

While it is unlikely the buy-side of the market will<br />

support the continued existence of three sets of<br />

standards with BPEL, <strong>BPM</strong>I and XPDL, neither<br />

alone offer a complete set of specifications for<br />

adequately describing an entire <strong>BPM</strong> solution.<br />

Lacking in <strong>BPM</strong>L, XPDL, and BPEL, for example, is<br />

a process modeling method capable of translating<br />

graphical models into executable process<br />

definitions. <strong>BPM</strong>I’s <strong>BPM</strong>N or “Business Process<br />

Modeling Notation” specification is designed for<br />

this purpose, to provide graphical notations for<br />

expressing business processes and is designed to<br />

translate these into <strong>BPM</strong>L or (eventually) BPEL.<br />

As specifications evolve and converge, a likely<br />

shake-out will be the engulfing of <strong>BPM</strong>L by BPEL.<br />

This leaves <strong>BPM</strong>N as a likely contender for the<br />

modeling requirements these two lack on their<br />

own. XDPL was designed as a process interchange<br />

standard and is currently used by some <strong>BPM</strong><br />

vendors for storing process definitions. But it<br />

lacks a graphical notion outside of <strong>BPM</strong>N (which<br />

is currently being mapped to XPDL). The<br />

advantages of XDPL over BPEL will inevitably lead<br />

to advancements in the latter, with the likely longterm<br />

scenario (greater than 12 months) being the<br />

co-existence of <strong>BPM</strong>N and BPEL.<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Fuego Product Assessment<br />

24<br />

Product Assessment: Fuego<br />

Overview<br />

Incorporated as a US-based company in 1999,<br />

Fuego was one of the first commercial software<br />

vendors to differentiate <strong>BPM</strong> as a unique<br />

discipline, distinct from EAI and applicationcentric<br />

workflow. Fuego was arguably the first to<br />

promote the concept of services orchestration over<br />

discrete automation and integration. Fuego is a<br />

privately held firm that has raised $45 Million in<br />

venture capital since its founding. Fuego is a<br />

independent software vendor focused on a single<br />

application suite, Fuego (formerly Fuego<strong>BPM</strong>).<br />

Orientation and Positioning<br />

Fuego is a <strong>BPM</strong> solution focused on "business<br />

services orchestration" or more broadly stated,<br />

the ability to develop a virtual catalog of<br />

componentized services from new and existing<br />

applications, and to define executable business<br />

processes able to invoke these services, then<br />

manage the execution and lifecycles of both<br />

processes and components. Architecturally, Fuego<br />

is not a single packaged application but a set of<br />

development, execution and management<br />

resources composed of six core modules:<br />

1. Process Designer – GUI-based, drag-and-drop<br />

process development environment, used to map<br />

processes, define business rules and edit syntax.<br />

2. Component Manager – server-side<br />

resource for developing business<br />

objects and managing library of<br />

application service components,<br />

presentations and integration<br />

points.<br />

3. Orchestration Engine – run-time<br />

environment for executing<br />

processes, business rules, and<br />

invoked application.<br />

4. Work Portal – J2EE-compliant environment for<br />

managing Work Lists (e.g., the process end user<br />

environment).<br />

5. Process Analyzer – OLAP reporting and analysis<br />

tool for analyzing process performance metrics;<br />

two cubes are pre-built: Process Workload and<br />

Process Performance<br />

6. Organization Administrator – server-side<br />

manager of roles, organizational structure and<br />

calendars.<br />

These modules, as well as their role in <strong>BPM</strong><br />

and services orchestration are described below.<br />

Process Modeling & Definition<br />

Processes are defined using the Fuego Process<br />

Designer, a GUI-based, drag-and-drop modeling<br />

environment with integrated scripting and<br />

debugging tools. Fuego presents a set of process<br />

design shapes that represent specific functionality<br />

(e.g., AUTOMATIC activity shape represents a<br />

process-to-system Activity; INTERACTIVE activity<br />

shape represents a process-to-human participant<br />

(GUI) Activity) but users can set preferences to use<br />

UML or user-defined icons with vertical or<br />

horizontal "swim lanes" used to identify roles that<br />

participate in a given process.<br />

Models can be developed using one set of shapes<br />

and "hot swapped" to another using a basic switch<br />

in the Process Designer model, or simultaneously<br />

developed in multiple modes by multiple<br />

designers. In this way, Process Designer is<br />

effectively methodology-agnostic in its visual<br />

metaphor. Process models are expressed in XML<br />

(using the XPDL schema) beneath the presentation<br />

layer, and compiled as executable Java code when<br />

published.<br />

Example Fuego Process Model<br />

Fuego’s process modeling tool technology<br />

partners, ProActivity and Proforma, have<br />

developed integration gateways which allow<br />

models in these environments to be translated into<br />

Fuego compliant models (specifically WfMC’s<br />

XPDL with Fuego Extensions). With existing<br />

models from these vendors, or UML models, there<br />

is insufficient syntax to define the level of service<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Fuego Product Assessment<br />

25<br />

interaction capability within Fuego. What they<br />

provide, however, is a means to leverage other<br />

process optimization initiatives or model designs.<br />

The third-party simulation and modeling tools<br />

also enable better integration of business concepts<br />

into the development process – by providing<br />

closed-loop reporting and analysis on process<br />

optimization.<br />

Fuego uses a Visual Basic like scripting language to<br />

define business rules for managing workflow,<br />

accessing, transforming and executing process<br />

objects, external applications, and data sources.<br />

Process Designer provides an IDE-type<br />

development environment for developing CIL<br />

instructions, and includes a syntax checker and an<br />

interactive debugger to test the business rules and<br />

services interactions.<br />

Process Designer supports collaborative design<br />

and development of process models and service<br />

interactions. This can be done hierarchically, with<br />

high-level flows designed by a business analyst<br />

and more granular service instructions developed<br />

by an application architect. This separation<br />

between business and technical functionality<br />

supports the objective of Fuego to enable<br />

processes to be defined by business users rather<br />

than programmers – a requirement reinforced by<br />

several data points in the survey detailed in this<br />

report.<br />

Development can also be done by a team of<br />

analysts simultaneously working on separate<br />

process elements. The security mechanism for<br />

controlling versions of subcomponents is handled<br />

by the LDAP security sign-in or with a third-party<br />

source code manager. Basic documentation is<br />

automatically generated and can be enhanced<br />

during process model definition in the Process<br />

Designer module.<br />

Process Execution<br />

Once defined within the Process Designer,<br />

processes are published as executable Java code to<br />

run and be managed by one or more of the Fuego<br />

Orchestration Engines. These Orchestration<br />

Engines provide the execution and management of<br />

all business process and business services accessed<br />

through the component manager (described later).<br />

Multiple Orchestration Engines can be deployed as<br />

a federation with a single engine designated to<br />

manage interaction between them.<br />

Communication between engines can be done via<br />

TCP/IP, as well as XML/SOAP over HTTP or HTTPS.<br />

Live Process Updates<br />

Live changes to process definitions or process<br />

parameters defined with the Fuego Organization<br />

Administrator (e.g., roles, calendars, etc) can be<br />

made on-the-fly without having to stop and restart<br />

a process instance. This enables the manner in<br />

which Fuego uses the notification mechanisms of<br />

an LDAP Server to poll for resources used by the<br />

Orchestration Engine when executing a process<br />

step.<br />

Fuego Work Portal<br />

Another point of separation between Fuego and<br />

alternative approaches for automation and<br />

integration is the ability to support manual<br />

activities and end-user participation. The Fuego<br />

Work Portal is the end user environment for<br />

engaging participants in process execution. The<br />

Work Portal generates unique work lists for each<br />

user. The Work Portal can be implemented to<br />

leverage LDAP or other directory services for user<br />

authentication, as well as extended to include<br />

external users (i.e., without named accounts) for<br />

B2B processes. The Work Portal can be assigned a<br />

URL for browser access or embedded in another<br />

web application.<br />

Another notable aspect of Fuego is the ability to<br />

define and maintain roles, organizational structure<br />

and calendars within the Fuego Organization<br />

Administrator. This can be used to enable rolebased<br />

routing of process objects, or combined with<br />

rules to enable escalation of process steps<br />

according to schedules and/or responsibilities.<br />

Process Integration<br />

What immediately differentiates Fuego from other<br />

process management approaches is the manner by<br />

which it enables application integration. Fuego<br />

employs a form of integration that is similar in<br />

approach to what is used by portal software<br />

vendors for building "portlets." Where it differs,<br />

however, is that the result contains not only<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Fuego Product Assessment<br />

26<br />

presentation logic (e.g., same as portlets), but also<br />

machine-readable instructions for how to invoke<br />

the component as an application service.<br />

Fuego's approach is distinct from the publish/<br />

subscribe model common to message-oriented<br />

integration middleware. Rather than relying on<br />

EAI infrastructure, such as connectors and<br />

message brokering, Fuego instead relies on the<br />

ability to define rules using its scripting language<br />

to describe where and how to access data and<br />

application services, then wraps these into<br />

individual components that are exposed as<br />

consumable services to process designers.<br />

This is accomplished by introspecting metadata<br />

from an API (Application Programmer Interface)<br />

to determine the interface points of the application<br />

for which access is sought. Introspection is an<br />

aspect of componentization that presents a<br />

standard means of generating self-describing<br />

software components. In the simplest sense,<br />

introspection is analogous to the recipe and<br />

ingredients list on a box of cake mix (the former<br />

representing "cake metadata"). In a software<br />

context, it is the practice of providing metadata<br />

that describes how to invoke an application<br />

through its API.<br />

In essence, introspection allows one application to<br />

read another's interfaces and data structures,<br />

without requiring explicit knowledge of how that<br />

application was deployed. The use of introspection<br />

is found in applications and APIs developed in<br />

recent versions of Java (since the release of JDK 1.2<br />

and the introduction of Java Foundation Classes)<br />

and is inherent in Enterprise Java Beans (EJBs), as<br />

well as other component architectures such as<br />

CORBA and COM/DCOM.<br />

In application environments with limited<br />

introspection, such as SAP's BAPI (Business<br />

Application Programming Interface), Visual Basic<br />

can be used or create a COM wrapper, or where<br />

available, the SAP DCOM Object Builder can be<br />

used. Outside of SAP, various other tools are<br />

available to enable introspection, yet in virtually<br />

all cases the process of exposing metadata is far<br />

simpler than building application-to-application<br />

integration. In a case where customer adapters<br />

are already in place, these provide sufficient<br />

introspection and can also be used. An example<br />

of Fuego being used to orchestrate SAP application<br />

services is illustrated in the case study at the end<br />

of this section.<br />

With Fuego, scripts are used within the Process<br />

Designer to define the syntax for accessing<br />

applications. Once a component has been<br />

cataloged using introspection to determine the<br />

structure of data tables, available data variables,<br />

input parameters, and the specific pieces of data or<br />

application functionality needed, during the<br />

creation of the process model the process designer<br />

uses the component needed to product some result<br />

– i.e., display a User Interface, access a data record.<br />

An example might be "Get Invoice" and as such the<br />

scripts instructions define where to find an<br />

invoice, which input variables are needed<br />

("Customer #", etc), and how the invoice appears<br />

when received. Once the instructions are defined<br />

and encapsulated as a component, the component<br />

is cataloged within the Fuego Component Manager<br />

as a resource for the Fuego Process Designer and<br />

ultimately the Fuego Orchestration Engine.<br />

This cycle (building components) is an essential<br />

element of the Fuego eco-system and its<br />

significance should not be lost in the technical<br />

granularity of its description. In short, the Fuego<br />

Component Manager presents a universal access<br />

point for virtualizing data and applications into<br />

consumable services. The Fuego Orchestration<br />

Engine orchestrates the sequence in which these<br />

services are utilized or consumed.<br />

This illustrates the essence of service orchestration<br />

versus integration and automation. While<br />

integration and automation are by-products of this<br />

process, the Fuego approach (of directly accessing<br />

application services and data) allows for optimal<br />

use of existing automation, with minimal<br />

investment in integration.<br />

Leveraging Web Services<br />

The component-oriented modules of Fuego<br />

(notably the Component Manager) allow for<br />

applications to be virtualized into service<br />

components. The same capabilities extend to<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Fuego Product Assessment<br />

27<br />

pre-existing components, such as Web services<br />

with WSDL descriptors, COM/DCOM objects, EJBs,<br />

Java classes, CORBA, XML, SOAP messages, or<br />

other self-describing software. These can also be<br />

registered or "cataloged" with the Component<br />

Manager as executable services. The degree of<br />

autonomy and self-description dictates the amount<br />

of effort and scripting required registering them.<br />

But once cataloged, each component is a service<br />

available to any Fuego process with a single<br />

method of invocation.<br />

Fuego allows for integration logic (how to find and<br />

invoke data and applications) to be abstracted and<br />

managed separately from process logic (how these<br />

services are used). This approach is in contrast<br />

with more traditional pairings of process<br />

management and EAI, where specific integration<br />

points are defined but not otherwise exposed as<br />

services. Rather than building integration logic<br />

into the process, the Component Manager provides<br />

a single method for accessing data and<br />

functionality from a broad set of systems and<br />

application types.<br />

In this sense, the Component Manager is<br />

orthogonal to the message bus used in publish/<br />

subscribe-type, message-oriented integration.<br />

Where these two part ways, however, is that the<br />

output of a message bus is meaningless to a<br />

process engine and serves only as a means for<br />

getting data from one point to another.<br />

In contrast, the Component Manager produces a<br />

result that is 100% consumable by the<br />

Orchestration Engine. Further, while the<br />

Component Manager is abstracted from the rest of<br />

the Fuego application modules, it is by design and<br />

definition far more tightly integrated than any<br />

third-party EAI software could be. It is looselycoupled<br />

with fine-grained interaction.<br />

Process Monitoring & Maintenance<br />

Executing Processes are managed by the<br />

Orchestration Engines. Inherent in the design of<br />

these engines are basic systems management<br />

capabilities, from back-up procedures and<br />

performance monitoring, to the ability to "listen"<br />

for execution problems and restart failed servers.<br />

A two-phase commit protocol is used to persist<br />

transactions and process state. In the event of<br />

failure, if the transaction was automated (e.g., an<br />

invoked application) it is rolled back to the<br />

previous state and restarted. If the transaction was<br />

a manual activity, the work item is sent back to the<br />

appropriate work list in the Work Portal.<br />

Fuego also provides a fail-safe mechanism through<br />

a system of one primary and multiple secondary<br />

Orchestration Engines. All Orchestration Engines<br />

are connected to one another notifying each other<br />

to check if they are active. If the primary<br />

Orchestration Engine stops working, one of the<br />

secondary Orchestration Engines will take the role<br />

of primary Orchestration Engine, managing the<br />

automatic and user generated transactions.<br />

Reporting<br />

Fuego Process Analyzer provides multidimensional<br />

OLAP reporting tools and analysis<br />

capabilities for analyzing process performance<br />

metrics captured by the Execution Engines.<br />

Extensions are provided for third-party report<br />

tools, such as Crystal Reports, however, there are<br />

comparatively sophisticated report design and<br />

analysis tools within Process Analyzer. Process<br />

Analyzer also allows for the creation of<br />

"dashboard" type live reports based on a discrete<br />

set of performance metrics. There are two<br />

predefined analysis cubes – Workload and Process<br />

Performance.<br />

Technology Platforms<br />

Fuego is a J2EE application framework requiring a<br />

Java Virtual Machine (JVM) to operate, as well as<br />

LDAP or SQL-compliant directory services, a Webserver<br />

(with a servlet container) or an application<br />

server, and a relational database. JVM (JDK 1.3 or<br />

higher) platforms supported including Windows<br />

NT or higher, as well as most major Unix variants<br />

(including Linux).<br />

Fuego supports any relational database with JDBC<br />

drivers, but already tested and verified are Oracle<br />

8i (Standard and Enterprise editions), Microsoft<br />

SQL Server (7.0 and higher), IBM DB2 (7.0 and<br />

higher), and Cloudscape.<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Tesoro Case Study<br />

28<br />

Standards Participation<br />

Fuego participates in various standards bodies and<br />

industry groups within the business process<br />

software space, including <strong>BPM</strong>I.org, ITAA, UDDI<br />

and RosettaNet. Both J2EE standards (JMS, JDBC,<br />

JNDI, JAAS, etc) and XML is used throughout the<br />

Fuego architecture, including process definitions<br />

that use the XPDL schema (the Workflow<br />

Management Coalition's XML Process Definition<br />

Language). Fuego is also exploring BPEL<br />

(discussed earlier in this report) as this<br />

specification evolves.<br />

Case Study: Tesoro Petroleum<br />

Highlights<br />

Orchestration of Existing Islands of Automation<br />

Distilling Complex Systems Into<br />

a Single Presentation Layer<br />

Reduced Training Requirements<br />

Improving the Accuracy, Timeliness,<br />

and Completeness of Data<br />

Improved Process Efficiency & Faster<br />

Inventory Turn-Over<br />

Cultural Shift From Packaged Software<br />

to Service Orchestration<br />

Deployment Environment<br />

Large-scale SAP deployment on Microsoft<br />

infrastructure.<br />

Case Study Overview<br />

Tesoro Petroleum is an $8 billion petroleum<br />

refining company in San Antonio, TX, using<br />

Fuego’s <strong>BPM</strong>S to address what Tesoro's CIO<br />

describes as the areas of "white space" across a<br />

major, multi-system SAP deployment. Tesoro<br />

estimates that close to 95% of its business runs on<br />

top of SAP. Fuego is used to acquire and post data<br />

to and from the SAP system, as well as to<br />

orchestrate user-centric processes that would be<br />

difficult (if not impossible) to define within SAP or<br />

through traditional EAI software.<br />

Problem Statement<br />

When the SAP system was built, Tesoro designed<br />

with a process-oriented architecture, rather than<br />

focusing on a single or set of software modules.<br />

This involved the development of a library of<br />

business scenarios and related documentation<br />

defining business processes. In this way, the SAP<br />

system covers the vast majority of the processes<br />

executed by Tesoro. What is left is the "white<br />

space" – the places in process requiring human<br />

intervention, as well as other disconnects between<br />

the flow of a business process and the specific<br />

application functions (i.e., SAP and other systems<br />

within a given process). Filling in this white space<br />

was the driver behind Tesoro's pursuit of a <strong>BPM</strong><br />

solution.<br />

Tesoro's problem was not one of automation but<br />

orchestration. Automation was already achieved<br />

through its significant investment in SAP. The<br />

challenge for Tesoro was to bridge the complex<br />

and distributed islands of automation through<br />

orchestration of discrete transactions within endto-end<br />

processes. A cultural challenge for Tesoro,<br />

was that after being an SAP shop for many years<br />

it was necessary to develop an application<br />

development discipline consistent with building<br />

the reusable components required to leverage<br />

orchestration strategy, rather than focusing on the<br />

configuration of packaged software (as is the case<br />

with SAP). Related to this was the need to develop<br />

a business analysis competency required to define<br />

end-to-end processes with enough granularity and<br />

validation to handle the necessary hand-offs<br />

between applications and users.<br />

Product Evaluation Process<br />

Tesoro required process management system able<br />

to deliver on two objectives:<br />

1. Define end-to-end processes that complement<br />

and extend the processes and automation<br />

already built within SAP (e.g., filling in the<br />

white space), and<br />

2. Allow for granular, yet intuitive definition of<br />

business processes such that they can be<br />

understood and refined by non-IT,<br />

business-oriented personnel.<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Tesoro Case Study<br />

29<br />

Tesoro began in 2001 with a survey of possible<br />

alternatives to address these issues. What they<br />

found, however, was that most options with the<br />

requisite integration capabilities were focused<br />

only on application-to-application transactions,<br />

and lacked the ability to define user-interfacing<br />

process steps.<br />

Consistent with the philosophy presented<br />

throughout this report, Tesoro understood the<br />

goal of <strong>BPM</strong> is to incorporate the people side<br />

of business processes with the application and<br />

systems side. Tesoro did look closely at leading<br />

integration platforms as a possible solution, but<br />

soon determined these were insufficient to<br />

address user-centric requirements.<br />

This led Tesoro to look beyond EAI solutions for<br />

an application that addressed the white space<br />

where individuals work outside of, or in concert,<br />

with other systems (rather than simply shuttling<br />

data between systems). Fuego was selected to<br />

create a process orchestration layer between users<br />

and systems. In this way Fuego provides the<br />

common presentation layer for users, then is able<br />

to parse-out and distribute pieces of a single user<br />

transaction and distributes the appropriate data to<br />

the three applications.<br />

Order-to-Cash<br />

Anything that has to do with inventory movements<br />

is a high priority for the oil refinery industry, and<br />

Tesoro is no exception to this. Oil is an expensive<br />

product to keep in inventory, and the goal of<br />

Tesoro is to move from the point of an order to the<br />

receipt of payment (read "cash") in the shortest<br />

time possible.<br />

This is what initially drove Tesoro to make such a<br />

sizable investment in SAP. Yet the inefficiencies<br />

remained at the points where discrete systems<br />

present hand-offs to users or other applications.<br />

Fuego is used to roll-up these discrete transactions<br />

across a given process (such as sales order entry)<br />

into a single transaction for SAP, ensuring the<br />

complete and most expedient entry of data, and<br />

negotiating the proper hand-offs between<br />

individual users.<br />

For any given sales transaction, the process<br />

involves creating a contract, then creating a sales<br />

order, then billing documents (payables and<br />

receivables), prior to payment. Across this process<br />

the same data needed to create a contract is<br />

required to be entered into three different systems<br />

(sales, goods, billing), which otherwise (prior to<br />

Fuego) requires three individuals to manually<br />

entered with each order.<br />

Areas of potential breakdown or inefficiency<br />

within this process include incomplete or<br />

erroneous data entered at any point, thus stalling<br />

the process or (worse) sending it back to be<br />

redone; inherent delays waiting for data to be<br />

entered; and the manual process of chasing down<br />

missing information.<br />

To resolve this, Tesoro built a portal interface with<br />

Fuego, which collects and validates all the<br />

information needed to complete a sales order, with<br />

each user seeing the environment appropriate for<br />

their own role, then rolls this up into a single<br />

transaction before it is posted to the appropriate<br />

SAP applications.<br />

Validation & Accuracy<br />

Service orchestration capabilities (such as those<br />

provide by Fuego) have opened the gates to far<br />

more and easier options for exposing and<br />

orchestration application functionality than was<br />

otherwise available with prior platform and EAI<br />

opportunities.<br />

Yet making use of these resources requires that<br />

the data driving these applications is precisely as<br />

is should be, or that the point of entry for data has<br />

the intelligence to resolve when this is not the case.<br />

Failing to resolve this presents a significant point<br />

of failure within any process management<br />

initiative. In Tesoro's case, it was determined that<br />

this difference represented 5 times the cost to<br />

make a correction to data within SAP than it does<br />

to enter it correctly the first time.<br />

Responding to the issues above, one of the first<br />

benefits Tesoro realized through its deployment<br />

of Fuego was to improve accuracy of both new<br />

and existing data posted to the SAP applications.<br />

©2003 Delphi Group • Ten Post Office Square, Boston, MA 02109-4603 • v(617) 247-1511 • www.delphigroup.com<br />

Reproduction authorization required.


Tesoro Case Study<br />

About This Document:<br />

The product-specific<br />

information contained in this<br />

document is intended to<br />

provide an overview of a<br />

specific product and vendor at<br />

the date of publishing. Facts<br />

presented have been verified to<br />

the best of our ability with the<br />

vendor and actual users of the<br />

product where indicated,<br />

however, Delphi cannot insure<br />

the accuracy of this<br />

information since products,<br />

vendors, and market conditions<br />

change rapidly. Delphi Group<br />

makes no implied or explicit<br />

warranties, endorsements, or<br />

recommendations in this report<br />

nor should such warranties be<br />

inferred from its contents. A<br />

complete assessment of your<br />

specific application, the method<br />

of implementation for a given<br />

product or technology, and the<br />

current state of that product<br />

must be considered in order for<br />

a recommendation to be made<br />

on any product’s suitability for<br />

your purpose, needs and<br />

requirements.<br />

Delphi Group is a leading<br />

provider of business and<br />

technology advisory services to<br />

Global 2000 organizations. With<br />

offices established around the<br />

world, Delphi has assisted<br />

professionals across disciplines<br />

and industries at nearly every<br />

major national and global<br />

organization and branch of<br />

government. Its clients and<br />

subscribers include more than<br />

half of the Global 2000.<br />

Delphi Group<br />

Ten Post Office Square<br />

Boston, MA 02109-4603<br />

v (617) 247.1511<br />

f (617) 247.4957<br />

Within the front-end of the Fuego environment (presentation layer)<br />

Tesoro defined rules and validation procedures, which while transparent<br />

to the user, allowed greater over the quality and accuracy of data entered<br />

or updated within the system.<br />

As part of the same, completeness is also improved by validation that<br />

all information for a particular transaction is present, and if not Fuego<br />

invokes an activity to either obtain it from the appropriate system or<br />

prompts the user to update it. Timeliness is also improved as Fuego<br />

facilitates more on-line data entry, rather than relying on faxes and<br />

manual entry.<br />

Componentization and Orchestration<br />

Tesoro used the SAP BAPI structures wrapped with COM (Common<br />

Object Model) to enable introspection. This, Tesoro reports, required<br />

about one month's development time, with another month spent<br />

developing the Fuego components and process definitions to orchestrate<br />

the SAP application services. Tesoro also used Fuego to authenticate<br />

SAP users using Active Directory (the directory services used internally).<br />

During this process Fuego was also used to create a common presentation<br />

layer that masks the complexity of the underlying SAP systems. This single<br />

environment replaced multiple SAP screens, reducing training costs, error<br />

rates and overall intimation from had been at times a bewildering SAP<br />

environment.<br />

Of equal significance, however, was that the use of Fuego enabled Tesoro<br />

to make a major architectural shift without changing its underlying<br />

infrastructure. What had been a single application tier with multiple<br />

accessing points and interfaces, was transformed into a three-layer<br />

architecture:<br />

Presentation Layer – a single means for access all systems through the<br />

Fuego Work Portal<br />

Orchestration Layer – the execution of process, validation and posting<br />

of data, and orchestration of servers, accomplished by Fuego and<br />

transparent to users<br />

Application Layer – Existing SAP systems and Microsoft infrastructure<br />

This presents an extensible infrastructure for deploying new processes<br />

and orchestrating additional services, without having to alter the user<br />

environment or back-end infrastructure. Looking forward, Tesoro is<br />

examining how they can address issues such as Sarbanes-Oxley<br />

Compliance with the Fuego <strong>BPM</strong>S. The incremental approach enabled by<br />

Fuego will allow for existing work to be leveraged in this process, such as<br />

monitoring SAP transactions for possible compliance violations.<br />

www.delphigroup.com

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!