About scheduling PolicyCenter processes
It is possible to schedule many, if not most, of the PolicyCenter processes for execution at periodic intervals. Guidewire recommends that you take the following information into account in setting up a processing schedule.
Processes business transactions accumulated at the end of specific business periods, such as business days, months, quarters, or years. |
|
Defers to periodic asynchronous background processing complex transactional processing triggered by user actions. |
Depending on your custom process, certain implementation details can vary.
The scheduler configuration file
It is possible to define your own schedule for a process. See Scheduling work queue writers and batch processes for more information.
Night-time processing
Night-time processing typically processes relatively large batches of work, such as processing premium payments that accumulate during the business day. Each type of night-time processing runs once during the night, when users are not active in the application.
Because most kinds of night-time processing typically operate on large batches and have
rigid processing windows, develop your night-time process as a custom work queue.
Night-time processing often requires processing items in parallel to achieve sufficient
throughput. Large workloads of nightly processing are typically too severe for the
servers in a PolicyCenter cluster that have the
batch server role.
For night-time batch processing, you typically configure a custom work queue table to segregate the work items of different work queues from each other. Each type of night-time processing typically has many worker tasks simultaneously accessing the queue for available work items. Using the same work queue table for all types of nightly work queues can degrade processing throughput due to table contention. In addition, segregating work items from different work queues into separate tables can ease recovery of failed work items before the nightly processing window closes.
Night-time processing frequently requires chaining so that completion of one type of batch processing starts another type of follow-on processing. Regardless of implementation mode – work queue or batch process – you most likely must develop custom process completion logic for your type of night-time processing.
Daytime processing
Typically, daytime processes are relatively small batches of work, such as reassigning activities escalated by users. Daytime processes run frequently during the day, while users are active in the application. You might schedule different types of daytime processes to run every hour or even every few minutes.
Even though daytime processing typically operates on small batches and lacks rigid processing
windows, develop your type of daytime batch processing as a custom work queue. Although
daytime processing often achieves sufficient throughput by processing items
sequentially, its workload on the servers with the batch server role
can slow overall performance of the application. As a work queue, worker tasks on
multiple servers (with the workqueue server role), can perform the work
in parallel.
If you develop your daytime processing as a custom work queue, you typically can use the standard work queue table for your work items. Different types of daytime processing run intermittently and in different bursts of relatively short work. So, table contention between various types of daytime batch processing often is minimal.
Daytime processes seldom requires chaining. Completion of one type of daytime process seldom starts another type of follow-on processing. Regardless of implementation mode, you most likely do not need to develop custom process completion logic for your type of daytime processing.