Measure Logic
Hello All,
Sorry if this has been asked before, but I can't find an answer.
Can someone explain the logic behind Measures (both numeric and non-numeric) and their Respective Thresholds (Upper Severe, Upper Warning, Lower Warning, Lower Severe)?
For example, I want to create a Filter Measure for a User Action SLA Business Transaction so that only PurePaths where there was a call to example.com/ExampleURI are returned, as well as track the Response Time inside of SLA constraints.
(Note: I know that for SLA BTs I'll have to create multiple BTs.)
I could go two ways.
1.) Create a single Web Request Response Time Measure as the Filter with "URI contains: /ExampleURI". What would I set Threshold at for <1s? How about for 1s-2s? Where would I set it? (U. Sever, U. Warning, etc.)
2.) Create 2 Filter Measures for each Business Transaction. A Web Request Count with "URI contains: /ExampleURI". What would I set the Threshold for that and under which of the 4 would I set it? Then, a Web Request Response Time Measure. What would I set the Threshold at for <1s? How about for 1s-2s? Under which of the 4 would I set it?
Thanks in advance.
Andreas,
Thanks for that explanation. It contained a lot of information. After doing quite a bit of research based on your response, things are much clearer.
One thing that really helped me out was an answer from our Guardian. He explained the following:
When using a non-numeric measure as a filter that includes options for Upper Severe, Upper Warning, Lower Warning, and Lower Severe, utilize Upper Sever as Greater Than or Equal To and use Lower Severe as Lower Than or Equal To.
Thanks once again for the answer.
So - non numeric measures such as, e.g: Method Arguments, User Agent, URI Pattern, ... are typically used as a Split Measure for your Business Transaction. This allows you to specify a Business Transaction that e.g: Filters on a specific URL (by using the Web Request Count Measure) and then SPLIT by the User Agent. The result would be that you get how many transactions you had on that URL by User Agent.
Having that said: non numeric measures are typically not used as filters. Why? Because comparing a non-numeric measure to a numeric threshold doesnt really make a whole lot of sense. The only thing where it makes sense is if you also change the evaluation option of these measure definitions to e.g: Occurence. In this case the Measure will not return the Alphanumer Value such as "Internet Explorer" but - the measure will return either 0 or 1 in case the String VAlue matches the Match Definition you specified. Now you can use it as a filter measure as you can specify a threshold of 1 which means: "DYnatrace - please consider all PurePaths where the Browser matches the name Internet Explorer as being part of that BT"
Here is an example of such a measure -> as you can also see. Everytime I define such measures I give it a meaningful name that tells me what this measure is really about. In this case this measure will tell me if this PurePath is generate from an Internet Explorer browser:
I hope this makes sense
Well, thanks for hosting them. They're really helpful.
A couple more doubts:
So, for non numeric measures, when should I use Upper Severe, Upper Warning, Lower Warning, or Lower Severe?
Also, for those same non numeric measures, does the aggregation matter?
Hi Cesar
First of all - thanks for the great questions during todays Live Q&A session. Happy to have folks on these webinars that are that eager to learn 
To answer your question I need to go explain a bit of the background on why we have these thresholds on measures.
When we first came out with dynatrace you could create measures and the thresholds were used to be displayed in the Charts (there is an option to display the threshold in a line chart) as well as for thresholds for Alerts (just as I've shown in todays webinar).
When we introduced Business Transactions we decided to "reuse" these thresholds for BT Filter Measures. Thats where some of the confusion comes in as in one scenario we use it for thresholds on charts and incidents - in the other case we use it to filter PurePaths.
Now - to your exampls
1) Yes - you could create that Web Request Response Time measure and set the static threshold for Severe and Warning and show this threshold int he chart or use it for your Incidents
2) For a BT that matches to those PurePaths that are calls to that URL you can create a Web Request Count Measure for that URL and use the upper severe threshold of 1. Why? Because dynatrace will evaluate every single PurePath. For those PurePaths where the URL matches this Count measure will return 1 - for those dont dont match it will return 0. Thats why all PurePaths where the Count measures returns >=1 will be tagged with that Business Transaction. Now the cool thing is that dynatrace automatically calcualtes Count, Response Time and Failure Rate of all the PurePaths that match to that BT. Not only can you chart this and also specify e.g: static threshold of the Response Time Measure of this Business Transaction -but you can also put this BT on the Application Overview so that dynatrace can do the automatic baselining.
I know - this is a long answer. Hope it makes sense
I suggest you check out some of the Business Transaction Links that you can find here: Dynatrace for Business
More on BTs will be discussed during our next Live Q&A session - scheduled for Dec 23rd - register here: http://bit.ly/onlineperfclinics
Andi
