Your data. Anywhere you go.

New Relic for iOS or Android


Download on the App Store    Android App on Google play


New Relic Insights App for iOS


Download on the App Store


Learn more

Close icon

What does this diagram mean?


#1

Hi,

I am not able to figure what is going on in my application, based on above 2 images. Here I am concerned about time period after 6.15.
I invoked my microservice from a client around 6.15.
First image shows application behaviour for first few mins. As can be seen, response time is shown as combination of JVM time(blue) and time consumed by external calls(green). Its in 4-5 ms range.
Second image shows application behaviour after first few minutes. As can be seen, there is a huge spike(previous bumps get dwarfed relatively), and all is attributed to JVM time.
What I know about my application is that external calls are quite heavy. But these images show JVM to be more expensive than external calls.
Also, on client side total time is shown as 398605 ms, and that is the approx time actually taken to get back the response. But this time is not same as in above 2 images.
What am I missing in my understanding here?


#2

Hey @manish. As you noted, this graph is showing that there is a lot of time being spent attributable to the JVM.

You’ll really want to correlate this information with what else is going on with the JVM and application. A few questions come to top of mind for troubleshooting. Is the JVM going into GC that correlates with this pattern? Are there an inordinate number of threads being created? What types of objects are being allocated when these requests come in? If requests are allocating large or a lot of objects, that would be worth investigating as well. Also, is there any logging data that correlates with the cycles you are seeing?
I would run through some of those scenarios and see if anything stands out.
Hope that helps.


#3

HI, @manish.ghildiyal: Another factor to keep in mind is that each data point on the chart represents the average of all transaction response times during one minute.

Your client is showing the response time of a single call (398605 ms); there may also be hundreds of other calls during that minute (to the same or other transactions) that are much faster, bringing down the average response time shown on the chart.


#4

I was the only one using this service during that specific time duration. So I am not sure if there are too many other calls to bring down average response time.
BTW, is there a way to check the response time for a specific call?


#5

Yes. Go to the Transactions page and select the desired transaction. The response time will be shown on the right-hand side: