07.02.2013 Views

Best Practices for SAP BI using DB2 9 for z/OS - IBM Redbooks

Best Practices for SAP BI using DB2 9 for z/OS - IBM Redbooks

Best Practices for SAP BI using DB2 9 for z/OS - IBM Redbooks

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

11.3.1 <strong>SAP</strong> <strong>BI</strong> database<br />

The four most relevant objects used <strong>for</strong> <strong>SAP</strong> <strong>BI</strong> (as described in Chapter 2,<br />

“Overview of <strong>SAP</strong> <strong>BI</strong> and related data” on page 9) are the master data, persistent<br />

staging area (PSA), data store (DSO), and InfoCubes (or simply cubes).<br />

For our Jupiter project, three database sizes were created and populated to<br />

supply the data volume requirements <strong>for</strong> the suite of tests. These are 5 TB,<br />

15 TB, and 25 TB.<br />

In reference to the <strong>BI</strong> database characteristics, it is important to note that:<br />

► The growth in database size was reached by increasing the size of the<br />

InfoCubes only. The data volume <strong>for</strong> PSA, DSO, and master data remained<br />

constant.<br />

► The data volume refers to cubes (F fact table and indexes) only. The actual<br />

<strong>SAP</strong> <strong>BI</strong> database size was larger, if we include the statistics <strong>for</strong> PSA, DSO,<br />

and master data.<br />

► <strong>DB2</strong> data compression was not turned on <strong>for</strong> this environment. Since the<br />

focus of the test was the per<strong>for</strong>mance scalability of the <strong>BI</strong> Accelerator, and<br />

since this feature is not available in all database plat<strong>for</strong>ms, <strong>SAP</strong> opted not to<br />

enable this <strong>DB2</strong> feature.<br />

11.3.2 <strong>BI</strong> database InfoCube load<br />

The initial build of the <strong>SAP</strong> <strong>BI</strong> database involved the population of master data,<br />

PSA, DSO, and InfoCubes. There were several challenges encountered during<br />

this phase of the study. Some were caused by the level of <strong>DB2</strong> code that we had<br />

installed at the time. Others were due to the characteristics of the database, such<br />

as the large volume of data being inserted and the concurrency of the jobs during<br />

data population.<br />

The execution of this project took six months. As standard benchmark practice,<br />

the software code level was not updated from the initial level unless we<br />

encountered a defect. However, due to the duration of this project, we did apply a<br />

few maintenance updates.<br />

The focus of the study was to showcase the per<strong>for</strong>mance scalability of the <strong>BI</strong><br />

Accelerator in various data volumes. However, the amount of time needed to<br />

load the InfoCubes was a significant portion of the project schedule. Thorough<br />

planning <strong>for</strong> disk storage allocation, layout, and backup/restore procedures was<br />

essential to meet the demanding milestones.<br />

Chapter 11. Project Jupiter: large <strong>BI</strong> Accelerator scalability evaluation 235

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!