App Overview with multiple BTs selected for same transaction
Hi
We have two BT for an application. 1) The build in Web Service by Method and 2) Transaction by Type (here Type is a method argument of the web methods)
All transaction in this applications is part of both BT.
We have added these two BTs in the app overview.
The question is will the app overview show same transaction twice as part of two BTs or not.
-Sreerag
The AppOverview (on the first tab) will show the aggregate of all selected Business Transactions for this App. If you have BTs that match the same PurePaths you will also get some of the timings calculated multiple times
Hi Andi,
I'm still trying to understand how the app overview works when we select more BTs and as you said the timings may be calculated multiple times.
For throughput, will the value double when i add another BT(that match all PPs) to the app overview? ideally it should remain the same no matter how many BTs you add in the app overview.
Assume that i have two BTs selected in app overview,
1) BT1 without any filter and applies for all purepaths for that application; failure rate 15% and RT 2s
2) BT2 with a filter and only 50% of the purepaths meet the filter criteria; failure rate 20% and RT 3s
Now if there are 100 PPs will the throughput be 100 (same as BT1 as it covers all PPs) or 150 (combined PPs of BT1 and BT2)
What will be the failure rate of the application? will it be 15% (same as BT1 as it covers all PPs) or 16.67% (combined failures of BT1 and BT2)
What will be the response time of the application? will it be 2s (same as BT1 as it covers all PPs) or 2.33s (combined RT of BT1 and BT2)
-Sreerag
Hi Sreerag
I can confirm that we use the SUM for the Throughput of those BTs that you have selected for the Application Overview. You App Overview will therefore show 150 (combined number of PPs).
For Failure Rate and Response Time we use a weight average based on the throughput of every BT.
Andi
Thanks Andi, this explains why users saw 2-3 times the throughput in app overview than they expected.
Throughput is OK as it doesn't trigger an alert, but don't the weight average on failure rate and response time skew the actual numbers?
-Sreerag
The weight average is only shown on the APpOverview. We do not alert on these as we alert on violations per BT and by BT Splitting. But - from a "display & visualization" perspective we weight it by also factoring in the load. So - if you have one BT that has 100% failure rate but very low throughtput and another one with 10% failure rate with high throughput the weight average is tending towards the 10% as this reflects the "real world"
hope this makes sense
Wouldn't it make more sense to just show the throughput, response time, and failure rate of the app? The current way seems kind of convoluted
Just had this cause some disruption at a customer of mine. Having a duplicated, triplicated, etc. throughput number makes no sense.
Engineering: please consider baselining the simple PurePath Response Time and failure rate measures by Application, rather than having the "throughput" of the application be the overlapping sigma of baselined BTs.
We back this request, It is always hard to explain how app overview works, especially when we on-board new applications into dynatrace. Also the values changes every time you add new BT into the app overview.
