Worker scalability

[SP_9431401]

I am using Windchill PDMLink Release 12.1 and Datecode with CPS 12.1.0.1

As business data to be processed in workers is dynamic, we need to check if we have reference of implementation of Autoscaling of workers for effective utilization of infrastructure.
or any other recommendation for better performance.

Here are the errors that I faced
there is delay in publishing of worker jobs

[VladimirN]

Article - "Windchill PDMLink: Publish Job delays occur when multiple Publisher Queue sets are configured with more numbered WVS Publisher Queues than workers in the mapped Worker Agent QUEUESETS": https://www.ptc.com/en/support/article/CS157233 

[vnamboodheri]

Hello @SP_9431401, 

It looks like you have a response from a community member. If it helped to answer your question please mark the reply as the Accepted Solution. 

Of course, if you have more to share on your issue, please let the Community know so other community members can continue to help you.

Thanks,
Vivek N.
Community Moderation Team.

[avillanueva]

From what  I can tell, workers run round robin. It will always check your first worker and then move on to others. The whole setup is scalable but I am not sure if that is dynamically scalable. I think what you are trying to achieve is scaling up publishing based on peak load and then dropping it down. First, let's dig into this deeper. Are you seeing performance issues currently? Are these related to long publishing times on a single jobs? Are these with large Creo assemblies and drawings? 

If yes to the last question, that's not a scale issue but you need to increase the speed of the worker. I had asked PTC a while ago if it was possible to kick these large jobs to a different queue or worker and the answer was no at the time. Remember that Creo is single threaded so make sure that you have a worker that is optimized for high single thread throughput. Most virtualized setups favor mutli-threaded throughput to switch between shared tasks.

To answer your other question, I have not seen a case where you can dynamically add more background MS or more workers as the queue increases. I am not sure what would be the point since workers will spin up as you need them based on the number of jobs.