Master Thesis - ICS

More documents

Recommendations

Info

Computer Science Department Antonis Misargopoulos Chapter 3. The High-Performance Scheduling Problem The computational Grids provide a platform for a new generation of applications [80]. Grid applications include ‘portable’ applications which can be executed on a number of computation and communication sites, resource-intensive applications which must aggregate distributed resources (memory, data, computation) to produce results. Grid applications include sequential, parallel and distributed programs. All of these applications can be executed on Grids simultaneously and share the available resources. Most importantly, each of these applications will seek to leverage the performance potential of computational Grids to optimize their own execution. From the application perspective, performance is the objective, while scheduling is fundamental to performance. Schedulers employ predictive models to evaluate the performance of the application on the underlying system, and use this information to determine an assignment of tasks, communication, and data to resources, with the goal of leveraging the performance potential of the target platform. As we saw earlier, in Grid environments applications share resources (computation resources, communication resources, instruments, data) and both applications and system components must be scheduled to achieve good performance. However, each scheduling mechanism may have a different performance goal. We can distinguish three kinds of schedulers: 1. Job schedulers promote the performance of the system, as measured by aggregate job performance, by optimizing throughput measured by the number of jobs executed by the system 2. Resource schedulers coordinate multiple requests for access to a given resource by optimizing fairness criteria to ensure that all requests are satisfied. Alternatively, they optimize resource utilization to measure the amount of the resource used. 3. Application schedulers, which promote the performance of individual applications on computational Grids by optimizing performance measures such as minimal execution, 22
Computer Science Department Antonis Misargopoulos speedup, or other application-centric cost measures. Note that both job schedulers and resource schedulers promote the performance of the system over the performance of the applications, as application schedulers do. In this setting, high-performance schedulers become a critical part of the programming environment for the computational Grid. However, high-performance scheduling on computational Grid is particularly challenging. Both software and hardware resources of the underlying system may exhibit heterogeneous performance characteristics; resources may be shared by other users; and networks, computers and data may exist in distinct administrative domains. In the following sections, we will describe what the term ‘application scheduling’ means and how a Grid scheduling model can be developed. 3.1 What is Application Scheduling? Grid applications consist of one or more tasks which may communicate and cooperate to form a single application. Scheduling Grid applications involves a number of activities [80].The most important of them are: 1. Select a set of resources on which to schedule the task(s) of the application. 2. Assign application task(s) to compute resources. 3. Distribute data or co-locate data and computation. 4. Order tasks on compute resources. 5. Order communication between tasks In the literature, item 1. is often termed resource location, resource selection, or resource discovery. Resource selection refers to the process of selecting candidate resources from a pool; resource discovery and resource location refer to the determination of which resources are available to the application. Item 2. may be called mapping, partitioning or placement. Also, note that items 1.-3. focus on the allocation of computation and data ‘in space’ and are termed generally as mapping; while items 4. and 5. are generally termed as scheduling and deal with the allocation of computation and communication ‘over time’. 23
Page 1 and 2: University of Crete Computer Scienc
Page 3 and 4: άκυκλους γράφους π
Page 5 and 6: the communication and the computati
Page 7 and 8: ΕΥΧΑΡΙΣΤΙΕΣ Αρχικά
Page 9 and 10: 6.1 Algorithm Description .........
Page 11 and 12: Figure 6.2: QuPGC Algorithm .......
Page 13 and 14: Computer Science Department Antonis
Page 33: Computer Science Department Antonis
Page 85 and 86:
Computer Science Department Antonis
Page 87 and 88:
Page 89 and 90:
Page 91 and 92:
Page 93 and 94:
Page 95 and 96:
Page 97 and 98:
Page 99 and 100:
Page 101 and 102:
Page 103 and 104:
Page 105 and 106:
Page 107 and 108:
Page 109 and 110:
Page 111 and 112:
Page 113 and 114:
Page 115 and 116:
Page 117 and 118:
Page 119 and 120:
Page 121 and 122:
Page 123 and 124:
Page 125 and 126:
Page 127 and 128:
Page 129 and 130:
Page 131 and 132:
Page 133 and 134:
Page 135 and 136:
Page 137 and 138:
Page 139 and 140:
Page 141 and 142:
Page 143 and 144:
Page 145 and 146:
Page 147 and 148:
Page 149 and 150:
Page 151 and 152:
Page 153 and 154:
Page 155 and 156:
Page 157 and 158:
Page 159 and 160:
Page 161 and 162:
Page 163 and 164:
Page 165 and 166:
Page 167 and 168:
Page 169 and 170:
Page 171 and 172:
Page 173 and 174:
Page 175 and 176:
Page 177 and 178:
Page 179 and 180:
Page 181 and 182:
Page 183 and 184:
Page 185 and 186:
show all

Master Thesis - ICS

Create successful ePaper yourself

Delete template?

Save as template?