Bookshelf Home | Contents | Index | PDF | ![]() ![]() |
Siebel Field Service Guide > Setting Up and Using Scheduling > Setting Up Server Key MappingsThis task is a step in Process of Setting Up Scheduling. Service regions are designed to be self-contained so that they can run on separate CPUs and even separate servers. Server key mappings are used to specify the location and routing of requests to these service regions. Server key mappings also assign the service regions to different processes, which determines whether they are processed in parallel or serially by the servers. There are two server key mappings, one for the ABS and one for the Optimizer. When the user clicks the Book Appointment button, the request is sent to the enterprise server. The enterprise server then looks at a map of the service regions and routes the request to the appropriate Siebel Server that stores that service region. Multiple Processor SupportIf the underlying operating system can support multiple processors on one computer, Siebel Scheduler can take advantage of those processors by running different service regions for the ABS and Optimizer on different processors. This is accomplished by setting different process numbers in a server key map. A server with a single processor can run multiple processes, but this may reduce performance. Service regions in different processes in different processes are run in parallel, while service regions in the same process are run serially. NOTE: The number of processes running on a server computer are not related to the number of processors in that computer. Because the ABS and Optimizer are different modules, they are already running as separate processes, even though the process number is the same. One module for one service region can run in only one process at a time For example, consider a four-processor computer, with only two of the processors in use. If one service region runs with both the ABS and Optimizer, the key map would look like the example shown in Table 30. Note that process refers to the number of processes for that component, not the processor number on the server. NOTE: You cannot run one module for a service region as two processes. If two service regions run on a four-processor computer, you can use all four processors, as in Table 31. Note that process refers to the number of processes for that component, not the processor number on the server. If there are more than two service regions on a four-processor machine, you can run them as different processes. For example, if the West Coast Region is very large and the Central and East Coast regions are smaller, the scheme shown in Table 32 might be the best use of the processors on a server. Note that process refers to the number of processes for that component, not the processor number on the server. The West Coast region, which requires more computation, uses two processors. The East Coast and Central regions run in the same processes and on the same two processors. Each server component is multithreaded and can load schedules for multiple service regions. These schedules are shared by multiple threads. Each request to the Optimizer runs on a new thread. A request identifies the schedule for one service region that it requires and locks it. Requests for schedules for other service regions can continue while requests for the same schedule are queued. As the response time for the ABS is quite fast, this should not result in high latency for the booking times being returned. Generally, this is true because the think time between ABS transactions is usually several minutes.
|
![]() |
![]() ![]() |
Siebel Field Service Guide |