QAID # 14010 Published
Question / Problem:
How do systems configured for Multiple Instances process Batches?
Answer / Solution:
All processing related to Capture modules (as opposed to KTM modules, for example), are performed at a Batch level.
When a Batch becomes available for a given module, the Batch Notification Service (BNS) sends out a broadcast letting other Capture machines know that a Batch is ready for processing. Listening nodes will process the Batch in a round robin manner.
For example, station A has completed scanning a Batch and it is ready for Recognition. The BNS broadcasts a message to other Capture nodes and station B, running mulitple instances of Recognition, receives the request. The first instance to receive the request will lock and process the Batch. All Documents within the Batch are processed by that instance and not distributed across the other instances. This allows the remaining instances to receive requests for processing.
This aspect of multiple instance support differs from KTM Server, where all Documents within the Batch can be ported across all of the services running on the node.
The Release module also selects Batches in the same way, however, only the first Release service will process Batches that have a Release Script not designated for multiple instance support. This mechanism ensures that a Release Script not designed for multiple instance support will not interact with resources (files) referenced or shared with another, asynchronous instance.