User Experience Monitoring

[soundarajan]

Hi,

Recently i've configured a User experience monitoring for one of my application. I'm able to see metrics getting reported for that on UI.

1). I've few doubts on dashboard data's, there is graph "Response time Distribution" i'm unable to understand this graph properly (attached snapshot), it shows some percentile check points.

I assume it shows how the end user response time got disbersed for selected time frame.

Could someone please advise on this.

2). Also once configured UEM , how to use it more efficiently and what sort of more value added monitoring data's we can get it from this.

3). Is it possible to extract specific user details who is accessing the application pages ? 

Regards,

Soundarajan

[Chitra.Lal]

Hi Soundarajan,

1). So this chart, "Response time Distribution", That part of EUM is tied to the Analytics. Basically the Controller pulls the EURT Distribution dataset from the Events Service Cluster.

- This metric, End User Response Time Distribution: If you are aware of End User Response Time metrics as described in the link, then this widget is distribution of just End User Response time over the Pages registered.

- Now a percentile is a measure that indicates a value below which a given percentage of values in a set falls: for example, the 99th percentile means that 99% of all values are below this level.

For more info, please refer to: https://docs.appdynamics.com/display/PRO44/Configure+Web+Percentile+Metrics 

2). You can use EUM data for a number of use cases and info about how your application is doing outside, from which part of the world, which are the problem areas on end user front etc and also the end user (your customers who are using your application) driven stats. EUM allows you to see how your web application is performing from the point of view of a real or synthetic end user.

- Apart from browser real monitoring, you can also use Browser Synthetic Monitoring, which uses geographically distributed Synthetic Agents to continuously test key user workflows in your application. This allows you to monitor the correctness and performance of multi-step flows independently of user-generated load.

I think the below doc links should be able to explain this better, please take a look:

https://docs.appdynamics.com/display/PRO44/Overview+of+End+User+Monitoring#OverviewofEndUserMonitori...

https://docs.appdynamics.com/display/PRO44/Browser+Real+User+Monitoring

https://docs.appdynamics.com/display/PRO44/Browser+Synthetic+Monitoring

3). You can by default see the user ip adress and geo info etc; but if you are specifically intersted in knowing the user name or email id of the user acceessing the page, you can do that by passing these info as custom user data in your browser snapshots and then use them as needed. 

Hope that clarifies. Do let me know in case you have any further queries.

Thank You,

Chitra

[soundarajan]

Hi Clal,

Thanks for your detailed results. 

I'm fine with my queries now.

One thing with the End user response time distribution is like im assuming that, response time distribution chart values should be correlated with the browser rum metrics breakdown (as in attached snapshot).

Could you please help on this alone.

Regards,

Soundarajan

[Chitra.Lal]

Hi Soundarajan,

Somehow I can't find any attachments in your posts so can'ts really see which screenhsot you are referring to, but to answer your query, response time distribution chart values should be correlated with the distribution of "End User Response time" metric for all the Pages registered in the time range selected.

Thanks,

Chitra

[soundarajan]

Hi Clal,

Thanks for your Patience & response, I'm pretty much clear on Response time distribution graph now.

I'll go through the graphs completely on this.

I've few queries on Network performance details, could you please help on this one.

I've attached word doc with details.

Regards,

Soundarajan

[Chitra.Lal]

Hi Soundarajan,

You understood it right, the server connection time value broadly comprises of below parameters

• Domain lookup time
• TCP Connect time
• SSL Handshake time

So if you want to monitor the complete network, look for the The Network time which should just be “server connection time” which is navigationStart to requestStart
navigationStart --> redirectStart --> redirectEnd --> fetchStart --> domainLookupStart --> domainLookupEnd --> connectStart --> (secureConnectStart) --> connectEnd

- Where as First byte time is Server connection time --> requestStart --> responseStart. That is navigationStart to responseStart.

- HTML Download Time is First byte time + responseEnd

Resource fetch time is calculated after HTML Download Time and is exclusive. It is 
domContentLoadedEventStart --> domContentLoadedEventEnd --> domComplete --> loadEventStart --> loadEventEnd

So the The total network time should be

EURT - Server Time - DOM Building time - On Load

Apart from this below are a few additional factors that can contribute to Server Connection Time:

1. Server Connection Time = navigationStart time to requestStart time (From NavTiming Capable Browser)
2. Apart from what is documented at https://docs.appdynamics.com/display/PRO44/Browser+RUM+Metrics, redirects and appCache can also add up to this duration.Please see the "Browser RUM Timing Metrics Overview" in the docs
3. These timings are reported by the Browser Navigation Timing APIs.You can find more fine details at

   http://www.w3.org/TR/navigation-timing/#processing-model

4. In general, variations would be based on network propagation times. 

Hope that clarifies. Let me know in case you have any further queries.

Thank You,

Chitra

[soundarajan]

Hi Chitra,

Thanks for your support, 

I've attached my understanding document of the End user response time distribution chart., Could you pls verify it once. It would be helpful.

Regards,

Soundarajan

[Chitra.Lal]

Hi Soundarajan,

You understood it right; A percentile is a value below which a given percentage of measurements in a set falls. For example, a 95th percentile value of 150 ms means that 95% of all values are 150 ms or less. Please refer to the attached doc link for better clarity:

https://docs.appdynamics.com/display/PRO44/Percentile+Metrics

Thanks,

Chitra

[soundarajan]

Hi Chitra,

Thanks for your continous support !!

While analysing the response slowness issue for an application across two locations, i'm bit puzzled on how to get to the prone zone.

I've an application running in Mumbai and Chennai location, in real time response value for Chennai is normal and for mumbai users experience much slowness.

So i tried to navigate it in UE to see how the things going there., but on the whole i got to see Avg response time for chennai is larger than Mumbai, im not able to pin down that slowness in UE, which mumbai user experiencing.

Could you pls suggest any way to work around this. Thanks

Regards,

Soundarajan

[Chitra.Lal]

Hi Soundarajan,

I think you have another thread open for this same query. Since we are handling it there already we'll request that we can close this thread here so that we do not get mixed up in the content discussing so many thing in teh same thread. 

I hope that is fine with you. Do let me know in case you have any concerns. 

Thank You,

Chitra