Updated to Fs-series but performance is poor.

Ville-Matti Linkoala 1

We have updated from B4ms to F4s_v2 with hopes to get more CPU time out of it. We are running webserver that is peaking in the old system to 100% frequently so plan was to change F4s for better performance. After we did that nothing really changed, still peaking 100% and when checking clock speed its only 2.6Ghz. I believe this should go to 3.7Ghz. Do you have any advice how to proceed?

singhh-msft 2,431 Reputation points

2021-03-08T08:39:01.63+00:00

@Ville-Matti Linkoala , thanks for reaching out to us. Happy to help.

To investigate further, can you tell how did you update from B4ms to F4s_v2 - any link that you referred?

Also, which factor was reaching 100% for your previous VM?

Further, please check if you are able to see F4s_v2 as your current size in your VM:
Ville-Matti Linkoala 1 Reputation point

2021-03-08T10:30:35.52+00:00

We just stopped VM(deallocate) and resized to new one and started again.
When I to to Insights and CPU utilization% is the factor that peaked 100%
singhh-msft 2,431 Reputation points

2021-03-08T11:03:53.55+00:00

Got it, in this case, I would advise you to run Performance Diagnostics on your VM and troubleshoot to find out the bottle necks: Azure VM Performance Diagnostics. I suppose, your old system is consuming more CPU than what F4s_v2 offers.

Also, pls keep track of Clock Speed: It features a sustained all core Turbo clock speed of 3.4 GHz and a maximum single-core turbo frequency of 3.7 GHz.

Let me know if this resolves your query.
Ville-Matti Linkoala 1 Reputation point

2021-03-08T11:28:40.3+00:00

I did run perfomance diagnostics on old system and CPU was one bottle neck. I dont know about those SQL issues because when monitoring Disk IOPS they are barely used at all so I would not think that those are the issues here.
singhh-msft 2,431 Reputation points

2021-03-08T12:36:03.43+00:00

Thanks for doing that, you can dig in more about the processes which caused these issues:

If you expand the Findings event, you'll see several key details. The tab lists the processes in descending order, per Average CPU consumption, and shows whether the process was related to the system, a Microsoft-owned app (SQL, IIS), or a third-party process.

This would make you aware about the processes which are causing high CPU usage.
singhh-msft 2,431 Reputation points

2021-03-08T12:37:02.797+00:00

To resolve this, you must figure out what is using the CPU. If your application is the culprit of increased CPU utilization, it could take considerable effort to debug and find the issue. In the meantime you can increase capacity temporarily while you look into the issue. If your VMs are receiving load-balanced traffic in an Availability Set, you can add more VMs to receive traffic, reducing the share of work each VM has to perform. Otherwise you can increase the size of your VMs to a type that has more vCPU.

For memory-intensive applications running on the JVM, high CPU usage can actually be a symptom of high memory usage. Check your JVM configuration, especially regarding the HEAP_SIZE and see if the CPU usage is caused by JVM garbage collection.

Let me know if this resolves your query.
Ville-Matti Linkoala 1 Reputation point

2021-03-08T14:31:08.933+00:00

I attach screenshot below about task manager and processes from this VM. This is the situation when only 1 user is logged in to webserver. When more users logs in same time it will peak to 100% and eventually become unusable. However I think we try next to resize to 8vcore CPU and see if that makes difference.
As you can see in the screenshot CPU speed is only 2.6Ghz. Shouldn't that be more like 3.7Ghz?
singhh-msft 2,431 Reputation points

2021-03-08T17:18:56.897+00:00

I see that your VM has Intel® Xeon® Platinum 8270 Processor, and it has 2.70 GHz Clock speed: Intel® Xeon® Platinum 8270 Processor. Let me reach out to PG regarding this discrepancy. Further, if you are okay, pls use higher vCPU VM and see if it helps. Otherwise, the code needs to be refactored to comply with the performance.

I would also suggest you to go through Azure App Service as an option which has automatic scaling and can help you in load balancing better.

Let me know if it answers your query.
singhh-msft 2,431 Reputation points

2021-03-09T05:08:14.383+00:00

I got a response from the PG that the f series base clock speed is 2.7GHz. It can boost up to 3.7 as needed to handle the load.

And since, the vCPUs are not changed after you updated the VM series (from B4ms to F4s_v2), the performance has not improved. If you choose a size which has more vCPUs than 4, you will see the difference.

Let me know if this answers your query.
singhh-msft 2,431 Reputation points

2021-03-09T11:09:17.987+00:00

@Ville-Matti Linkoala , just checking in to see if the above comments help your query.
singhh-msft 2,431 Reputation points

2021-03-10T05:34:29.763+00:00

@Ville-Matti Linkoala , just checking in to see if the below answer helped you.
singhh-msft 2,431 Reputation points

2021-03-10T16:29:08.487+00:00

@Ville-Matti Linkoala , just checking in to see if the below answer helped you.

1 answer

singhh-msft 2,431 Reputation points

2021-03-09T12:07:43.367+00:00

@Ville-Matti Linkoala ,

After we did that nothing really changed, still peaking 100% and when checking clock speed its only 2.6Ghz.

The f series base clock speed is 2.7GHz. It can boost up to 3.7GHz as needed to handle the load. And since, the vCPUs are not changed after you updated the VM series (from B4ms to F4s_v2), the performance has not improved. If you choose a size which has more vCPUs than 4, you will see the difference.

To resolve performance issues, in general, you must figure out what is using the CPU. If your application is the culprit of increased CPU utilization, it could take considerable effort to debug and find the issue. In the meantime you can increase capacity temporarily while you look into the issue. If your VMs are receiving load-balanced traffic in an Availability Set, you can add more VMs to receive traffic, reducing the share of work each VM has to perform. Otherwise you can increase the size of your VMs to a type that has more vCPU.

For memory-intensive applications running on the JVM, high CPU usage can actually be a symptom of high memory usage. Check your JVM configuration, especially regarding the HEAP_SIZE and see if the CPU usage is caused by JVM garbage collection.

Further, I would advise you to run Performance Diagnostics on your VM and troubleshoot to find out the bottle necks: Azure VM Performance Diagnostics. Using this, you can dig in more about the processes which caused these issues:

If you expand the Findings event, you'll see several key details. The tab lists the processes in descending order, per Average CPU consumption, and shows whether the process was related to the system, a Microsoft-owned app (SQL, IIS), or a third-party process.

This would make you aware about the processes which are causing high CPU usage and will help you remove those bottlenecks.

Let me know if this answers your question.

-----------------

Please do not forget to "Accept the answer" wherever the information provided helps you. This will help others in the community as well.
Please sign in to rate this answer.

0 comments No comments
Sign in to comment

Use comments to ask for clarification, additional information, or improvements to the question.

Share via

Updated to Fs-series but performance is poor.

1 answer

Your answer