Processing modes

PolicyCenter supports two modes of processing:

• Work queue
• Batch process

Work queue

A work queue operates on a batch of items in parallel. PolicyCenter distributes work queues across all servers in a PolicyCenter cluster that have the appropriate role. In the base configuration, Guidewire assigns this functionality to the workqueue server role.

A work queue comprises the following components:

• A processing thread, known as a writer, that selects a group (batch) of business records to process. For each business record (a policy record, for example), the writer creates an associated work item.
• A queue of selected work items.
• One or more tasks, known as workers, that process the individual work items to completion. Each worker is a short-lived task that exists in a thread pool. Each work queue on a cluster member shares the same thread pool. By default, each work queue starts a single worker on each server with the appropriate role, unless configured otherwise.

Work queues are suitable for high volume batch processing that requires the parallel processing of items to achieve an acceptable throughput rate.

Batch process

A batch process operates on a batch of items sequentially. Batch processes are suitable for low volume batch processing that achieves an acceptable throughput rate as the batch process processes items in sequence. For example, writers for work queues operate as batch processes because they can select items for a batch and write them to their work queues relatively quickly.