1

I understand that my question is not really specific and that it has no clear answer, but I need help to figure out where to start looking.

So we work on portal azure and have a web job that takes messages from a queue, and processes them; for some weird reason when I use the S1 plan (specs: 1x cores, 100 total ACU -I think its some unit for azure: azure compute unit- 1.75 GB memory, A-Series compute) each message takes around 1 minute to process, but when I upgrade the S3 plan (specs: 4x cores, 400 total ACU, 7 GB memory, A-Series compute) each message takes 3 minutes... I would understand if the each message would take the same amount of time as it is being processed linearly, and maybe increasing the processor might not really have a big effect on it, but how could it possibly make it slower?

I would appreciate any suggestion to help me debug this problem.

More info

The web app is a visual studio MVC project.

A.J Alhorr
  • 527
  • 7
  • 27

1 Answers1

1

The time taken for processing the messages will considerably improve if the tier is increased. So there might be something wrong with the logic which processes the message. Please revisit the logic.

In case of Web jobs, you need to increase the tier of the App Service to scale it up. As the App Service is upgraded to a higher tier, the memory and the CPU are shared between the App Service and the Web Job. You cannot scale the Web job without scaling the App Service.

My suggestion is to replace Web jobs to Cloud Services (Worker role). I had Web jobs for processing the messages from Queue, it turned out to be inefficient in case of scaling and Azure don't suggest Web Jobs any more. So I replaced it with Worker role. You can also try Azure Functions to process the message.

Arunprabhu
  • 1,454
  • 9
  • 15
  • Thank you for your response, I will check this "worker role"; and currently we actually publish the web job on a separate plan so that the API and webservice don't share resources. – A.J Alhorr Oct 02 '18 at 07:35