Dynamic Workflow Application Scheduling with Quality of Service

Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 4.5 Modeling Additional Dependencies or Constraints . . . . . . . . . . . . . . 117

4.5.1 Discounts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 4.5.2 Synchronization Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 4.5.3 Associated Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 4.5.4 Integer Data Sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 4.5.5 Data Sizes Known at Runtime Only . . . . . . . . . . . . . . . . . . . . 118 4.5.6 Data Servers or Repositories . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119

4.6 Checkpointing in Workflow Optimization . . . . . . . . . . . . . . . . . . . . . . . 120

of

The quality metrics identified and proposed in Section 3.1 need to be gathered for services, their providers, and the network that separates the client and the service. Moreover, in many cases the raw measured values should not be a basis for the values used in workflow application scheduling due to the following reasons:

1. Some values may be short-lived and may not reflect the actual values over a period of time.