Troubleshooting VM Performance

Question

So, I'm trying to troubleshoot some app performance issues for a client. They say the app is slow at times, and things that would take a couple of seconds now take minutes to perform.

My question is, why is there a difference between metrics and insights? Insights is showing a 20% decrease in utilization... so who is right here?

I read that Insights is like peeking at the Task Manager locally (which, in my opinion, would be more accurate) than Metrics looking at the host as a whole.

My dilemma is... if the client is complaining about performance and I am seeing higher usage up to 80-90% in Metrics but Insights is showing 50-30%, what do I do?

Accurately speaking, IMO Insights is correct, but Metrics could be telling of something... again just confused.

This is a D2as_v5 (2C/8GB) the minimum required specs for the app. Only two users are using it in a client-server setup. Memory hovers at 4GBs usage no matter what and has always been stable even during these peak CPU usage. We are using P10 SSDs.

Answer 1

Hi,

When it comes to performance issue, its important that you understand how the application works and then try to narrow down by doing some troubleshooting to see which component is at fault.

In a generic manner, what you checked for is absolutely right. Checking for spike in CPU, Memory and higher IOPS, etc.

Is this a web site ? Where is the client located ? How is the client trying to access this site ? Is it over private network or Internet ? When other client from different destination tries to access the site, was it working ? Is it an issue from client end or at the server end? Can you try to deploy the same application with a higher SKU and check if there are any improvements ?

Its more as asking right questions to narrow down the issue and then take corrective action to fix it.

Regards,

Karthik Srinivas