Performance tips

Message queue performance considerations

Message Queuing (MQ) guarantees delivery of messages but does not guarantee delivery timelines. Therefore, MQ is not recommended for time-sensitive interactions such as UI updates or workflow synchronization.

As the number of active MQ activities increases, the processing time on the Work Tasks engine also increases. For optimal performance, it is recommended not to have more than 1000 MQ activities in a waiting state at any given time.

Recommended steps for optimum performance

1. AVEVA Work Tasks dlls configured in the SKAddInProviders table and AVEVA Work Tasks table schemas can be cached for better performance.

   • The code given below needs to be put in the application on start of the web application.

     Skelta.Core.EnableCaching.Initialize(new  Skelta.Core.ApplicationObject("ApplicationName"), null);

   • If new dlls are added/existing dlls are changed and if they need to be re-cached the below code can be used.

     Skelta.Core.EnableCaching.Reset(new Skelta.Core.ApplicationObject("ApplicationName"), null);

   • For resetting the cache in engine, use the code given below. By default when the engine starts all the dlls and table schemas are cached.

     Workflow.NET.Engine.Client client = new Workflow.NET.Engine.Client("ApplicationName","WorkflowName"); client.ResetCache();client.Close();

2. You can set the threads that can be used by AVEVA Work Tasks in the farm configuration UI.

   • AVEVA Work Tasks Engine properties have the option to set Thread Pool Size.

3. You can ensure reliable performance even under heavy loads.

   • To ensure reliable performance even under heavy loads make sure Connection Pooling is enabled. By default, pooling is enabled and Max Pool Size is 300.

4. Go for fail over clustering if the loads expected are high. Use different servers for running the AVEVA Work Tasks Service and Databases.