I've been trying to get a particular dashboard to work which displays the server response time and the perceived render time for steps over an online sales funnel.
The issue is I cannot seem to create a chart based on the perceived render time for the 'landing' page only. To be clear, I'd like to see a dual chart with: server response time and perceived load time for 1 page only. Though, I'd repeat this type of chart for each step in the sales funnel.
This can be set-up for the response time, by adding a web request count measure against the specific uri and adding it to a business transaction. e.g. measure for 'new user' = count web request on uri '/new_user.jsp'. The splittings for this work as expected.
However, when trying to do it against UEM/User Actions. e.g. measure for 'new user' = count user action on actions ending on uri '/new_user.jsp'. The spittings pick up lots of different items.
I've tried, DOM, perceived load, user action count etc. start and end on page, but each of these business transactions are picking up sub-requests in the splittings and adding overall noise where I can't pinpoint the one page.
It's not until drilling down to the user action purepath on the BT, that the listing looks correct, e.g. they all say: Loading of '/new_user.jsp'
Thanks in advance for any help.
Solved! Go to Solution.
I would suggest the following Business transaction:
This configuration will return the server contribution and the perceived render time per loaded page. Then, you can chart the resulting measures on a dual axis chart (one for each loaded page). I would recommend to use the single chart as you can use the splitting options and automatically create a chart per loaded page.
To explain the behavior you noticed, I would need to know which type of Business Transaction you created and what the splitting configuration was. I think that in the first case, you filtered on Web Request, which is equivalent to PurePaths (if you start PurePaths on Web Request). So you have only the corresponding calls. In the second case, you probably filtered on User Action for a PurePath Business Transactions, which would return all the PurePaths associated to a specific User Action (the noise you mentioned).
Please let me know if I answered your question.
Thanks for your reply Nicolas.
The main item that was the issue was the splitting. Basically, IMO I now use this convention for a 'must-have' in these BTs, or at least a filter based upon them as explained below:
This can be augmented with other splitting types (filters) against the base measure, i.e. getting the actual page-name, using regex, i.e. for a user action URL, you can create a transform filter for that particular measure (/folder/"pagename".jsp) by editing the original AND saving a custom version.
This helped a lot with understanding where the measure came from and stopped the blank or abstract splittings - as some splittings really don't work with certain BTs, though it's not immediately obvious. You pretty much picked my test attempts as using different measure types and different splitting to find the sweet spot. A bit of a learning curve for sure.
Trying to measure the perceived (and other loading times). I was able to achieve this by using: