Overview of Automation Services and Load Balancing

Overview

The Automation Service in PSIcapture allows for automated processing of documents in two ways: Auto Import and Workflow Auto Processing.

Auto Import is used to automatically import documents.The service monitors locations for incoming files, which are then collected into queue directories for processing. When a queue directory is created, a reference to the directory is stored internally in PSIcapture. The Automation service works through this list of queue directories one at a time, generating a new batch for each, and then importing and processing the contents of the directory.

Workflow Auto Processing (WAP) is used to automatically process workflow steps for existing, checked-in batches in the central Batch Manager. This allows the PSIcapture Automation Service to detect and complete these batches without additional human intervention.

Spreading the Auto Import Load

PSIcapture is installed in one of three ways: a stand-alone mode, a server mode, or a networked mode working in concert with a PSIcapture server. Stand-alone and server installations handle the duties of monitoring, queuing files, and maintaining the list of queue directory references. They can also create and process the resulting batches.

Machines with network installations can be called upon by the server to create and process batches as well when the server is busy, as shown in Figure 1 below. The more network stations you have configured for Auto Import, the more batches your environment can process simultaneously.

Figure 1
Figure 1 - Classic auto import with a remote workstation

Spreading the Auto Import Load – Alternative Method

Although less common, it is also possible to use multiple PSIcapture servers or stand-alone installations to spread auto import loads (Figure 2). Each must have its own set of capture profiles and separate monitored areas – multiple machines must never monitor the same area. The maintenance difficulties inherent with such an arrangement make this a less desirable option.

Because a stand-alone PSIcapture installation cannot interact with other PSIcapture machines, it is unable to share its auto import load with other installations.


Figure 2 - Load sharing among multiple PSIcapture servers.

Workflow Auto Processing

PSIcapture version 5.0 introduced an automation enhancement called Workflow Auto Processing (WAP). When enabled, each PSIcapture workstation can be tailored to automatically process workflow steps for existing, checked-in batches in the central Batch Manager. This allows the PSIcapture Automation Service to detect and complete these batches without additional human intervention.


Figure 3 - Details of workflow auto processing.

Machines are configured for WAP in one of two modes. The first mode completes batches whose capture profile is set to auto-process through the entire workflow (Attempt to Auto Process Batch through Workflow checkbox in the Workflow section of capture profile configuration). If a batch stops processing due to an exception or reaches a manual step (such as Quality Assurance), this mode is designed to detect when such batches are ready to proceed with the next workflow step, and then resubmits them for auto-processing.

The second mode works with batches whose capture profiles are not set to auto-process batches through the entire workflow. Each machine in a server/network configuration can be configured to complete one or more workflow steps for these batches. For example, machine #1 can be configured to process OCR workflow steps, while machine #2 can be reserved for Indexing steps.

Machines can be configured exclusively for WAP, exclusively for auto import, or for both WAP and auto import. When configured for both WAP and auto import, WAP takes precedence.

Load Sharing Considerations with Workflow Auto Processing

Workflow Auto Processing (WAP) can have either a positive or a negative effect on overall load sharing. A machine can either process a queued set of files for import or process a batch through Workflow Auto Processing; it can’t do both simultaneously.

Because WAP takes precedence over import, having WAP enabled on every machine in your networked environment has the potential to slow down the normal auto import process. If you have hundreds of batches waiting for completion in the Batch Manager, the available automation workers will be consumed with WAP duties. This leaves no workers available for importing files. New files will continue to queue, but none of the queue directories will be processed into batches until workers become available.

Having too few WAP stations can have a similar affect on overall batch completion. Files in queue directories will be quickly converted to batches. However, incomplete batches will build up in the Batch Manager, requiring more human intervention and thus longer completion times.

When using Workflow Auto Processing, make sure you have the proper mix of auto import and WAP workstations in your environment to ensure smooth operations.

Licensing Considerations

When planning your distributed PSIcapture environment, you must consider the number of licenses you will need to support your operations. PSIcapture consumes licenses as follows:

  • Each PSIcapture server and network station that has the Automation Service enabled will consume a single license in your License Manager, regardless of configuration.
  • Network stations that are used solely for client interaction with PSIcapture through the GUI will also consume a license when the GUI is started.
  • If a user opens the GUI on a network station that has automation enabled, only one license is consumed. It is shared by both the client and the Automation Server.

The Automation Service

The diagram below provides an overall view of the entire Automation Service encompassing both Auto Import and Workflow Auto Processing.

  • The PSIcapture server’s Automation Service handles all duties related to directory monitoring and file queuing.
  • Network stations can import files or process batches assigned by the PSIcapture server.
  • Each network station that performs imports must have full access to the monitored areas where queued directories are stored.
  • Each network station registers itself with the PSIcapture server’s Networking Monitor at startup. The Networking Monitor maintains a Worker List of available network stations that can handle extra Auto Import and/or Workflow Auto Processing duties for the server.
  • The Directory Monitor watches for incoming files in the monitored directories defined in the PSIcapture server’s capture profile. When new files are detected, the files are moved to a new queue folder. A reference to this queue folder is stored in the Queued Folders List.
  • The Workflow Auto Processing/Queue Monitor is in charge of assigning Workflow Auto Processing and Auto Import work to the next available worker in the Worker List.
  • Once a worker completes its assigned task, it reports its results back to the server and is made available again for additional work.


If you have further questions, please contact our Support team by: