Monitoring Azure managed Instance Metrics

Question

Hi,

Which is the recommended solution to monitor Azure managed Instance metrics. Performance stats for items like some of the below.

Buffer Manager: BufferCache Hit-Ratio
Buffer Manager: Page Life Expectancy
General Statistics: Logical Connections
General Statistics: User Connections

I use currently use nagios to monitor and graph so that I can review these metrics over days/weeks/months etc for our onsite managed instance and pin point any issues. I thought I had seen similar type graphs/charts in Azure as well, but I can't find them. Would you use Azure monitor etc to do this?

Thanks

Answer 1

Here you will find how to collect performance counters on Azure Managed Instance.

Answer 2

For tracking CPU usage, my suggestion is to use the following counters:

• Processor: % Privileged Time
• Processor: %User Time
• System: Processor Queue Length

For monitoring Storage Subsystem, use the performance counters outlined on this article.

For monitoring memory usage, use the performance counters outlined on this article.

Answer 3

@Alberto Morillo Thank you for your reply. I already know the performance counters that I want access to. I already collect them using nagios from our VM servers running sql. Most of the performance counters you listed appear to be dependent on access to the windows server layer. However, been an Azure managed instance we don't have access to these server counters as far as I know.

Also the sql queries will only give you stats when you run them at that time. I'm looking for a mechanism that allow constant polling of counters and then graphing them for Azure managed instance like our nagios platform allows us to do for our sql instances. It's really very helpful to troubleshoot issues when you have a historical view of metrics for the server and the respective sql instance running on that server.

There are very limited metrics in graph form for the managed instance as listed below from Managed instance -> Metrics

Average CPU percentage
IO Bytes Read
IO Bytes Written
IO Requests Count
Storage Space Reserved
Storage Space Used
Virtual Core Count

However, none of these gives you actual stats on how the instance itself is doing health wise. There must be a way to get poll these stats and graph them.