Diagnostics additional information

[j82w]

The current diagnostics can have a lot of pages. This issue is to describe adding a summary to the top the diagnostics to make it easier to root cause.

Example of suggestion:

"TotalRequestStats":{
   "TotalRequests": 2,
   "FailedRequests":0,
   "ThrottledRequests":0,
   "MaxCpuUsage": 25,
   "TotalTimeInMs": 50,
   "TotalServiceProcessingTimeInMs": 30,
   "TotalNetworkingTimeInMs": 40,
   "TotalSdkTimeInMs":10
}

[sourabh1007]

What is the difference between TotalServiceProcessingTimeInMs and TotalNetworkingTimeInMs.

If we all feel like that this information is essential then we can discuss about adding this information in telemetry also.

[j82w]

TotalServiceProcessingTimeInMs is the time it took for Cosmos DB backend to process the request. It does not include any networking time like how long it was in the queue waiting to be processed. It's essentially RntbdRequestStats.TransitTime - response.BELatencyInMs

TotalNetworkingTime is the time from when the SDK put the request on the wire and until it got a response.

[ealsur]

Adding a summary would greatly simplify the areas where customer can laser-focus on.

[kirankumarkolli]

How about include the wait-times? (concurrency forced waiting times)

[j82w]

@kirankumarkolli can you expand on what you mean by concurrency forced waiting times?

[kirankumarkolli]

@asketagarwal what else is pending?

[asketagarwal]

#2732