Most widely used new relic metrics for performance/load test on front-end and backend

Hi,

What are some of the most common dashboard metrics within insights used by new relic for performance and load? I need some standards metrics to capture and how to query them for both front and backend for tracking

This seems like a question for @stefan_garnham and @eyedean!

Got any extra helpful queries to share besides what is in our Insights NRQL Queries Library post?

It’s a very broad question. Usually NR questions come as “I want to track X, how can I do that in NR?” but now it is “What are some good X’s that I should track?” Interesting! :slight_smile:


I’d say for front-end:

  • Throughput on key pages (you can use TIMESERIES COMPARE WITH 1 WEEKS AGO on it)
  • Average and Median of Page Load Time (especially for Initial Landing Pages solely). Note that Median is way more important than Average in this case due to outliers – I could talk for 40 minutes about it in last year’s FutureStack 2016! I usually add 50th, 75th and 90th percentile (in the same widget) on anything related to client-side timing using FILTER.
  • Break down the above two for different device categories (Desktop vs Mobile) on different charts.
  • Also break it down by pages on your website like login, homepage, authenticated page, profile, etc.
  • [Advanced] Assuming you have SPA (Pro) you can dive into BrowserInteraction attributes to see what percentage of the load time delay is for generating initial HTML (up to responseEnd event) and what percentage of that is for too many scripts and codes in front-end.

So you should have 2 charts (each showing at least two TIMESERIES line) x 2 categories x 3 page-types = 12 charts here. That’s enough for a good front-end dashboard!

On back-end:

  • Throughput and response time of top transactions (authentication, search, etc) assuming you have a RESTful archtecture.
  • Breakdown of calls per HTTPResponseCode like 200 vs 404 (you can have all of them in one-chart)
  • Success rate in terms of your own business logic on a call. It might require adding a custom_attribute to your Transactions. Basically, you should be able to see what percentage of, for example login, calls are successful over time.
  • DAU (Daily Active Users) if you have a unique user id that you can track.

Oh, and I am not going to spoil the exercise and share the queries right away! It’s a great learning opportunity to be walked by foot instead of taking the airplane. :wink:

1 Like

Cannot fault any of @eyedean’s items and they are a good start.

I have found it useful to use the rate function to look at error rates on transactions or page load rates to determine busy periods during the day, week and month.

For management reporting, I use Synthetic test results to calculate the availability of the sites along with page load times for the last month compared with the previous month to show any increase or decrease in performance.

If you discover any additional useful queries as you are working your dashboards @amaddox, then do add them to the Insights NRQL Queries Library post. I created it as an accelerated learning resource for the community.

2 Likes