By Splunk April 12, 2016
New Relic is a popular Application Performance Monitoring (APM) solution that developers use to monitor and diagnose the performance of their code. SignalFx is integrated with New Relic so that our joint customers can get more out of their APM metrics, bringing them into the context of behavior and performance from every part of the stack. New Relic customers who connect to SignalFx can take advantage of SignalFx’s real-time analytics, alert detectors and broader view of the environment.
To get started with New Relic metrics in your SignalFx organization, just select the New Relic Integration from the Integrations tab.
Add your New Relic API token (or multiple tokens if you have multiple accounts that you’d like to get metrics from) and connect. You’ll see New Relic metrics appear in SignalFx within minutes.
To pull in data from New Relic’s API, we first obtain the list of metrics with available data points, then make a second API call to get available metric values. New Relic recently upgraded their API to make the results from these two endpoints more consistent, making the results much smoother.
Next, we make a few changes to the metric names that come back from the API. Metric names from different New Relic modules are constructed using different patterns, so they can’t be reliably transformed into dimensions. Instead, New Relic customers in SignalFx can use components of metric names as dimensions on-the-fly while building charts. We call this capability “node aliasing“. To make this as smooth as possible, we ensure that all the information that you might use to filter or aggregate your data is included in the name of the metric itself. To the end of each metric name we add the application or server ID that it’s describing, and to the beginning of each Servers metric name we add its account ID.
We created a chart-building mode specifically for New Relic data that lets New Relic customers take full advantage of SignalFx dimensions, filters and analytics. Called “New Relic wildcard mode”, this feature performs wildcard searches for metrics delimited by the slash “/” character. For example, the Apdex score for application 12742967 is given by a metric that looks something like this:
Apdex/score/12742967
To see the Apdex scores for that endpoint in all applications, search using New Relic wildcard mode like this:
Apdex/score/*
Read more about New Relic wildcarding mode here.
Best of all, if you install the SignalFx collectd agent on the same servers that host your New Relic-instrumented applications, you’ll be able to connect New Relic’s stellar applications data with SignalFx’s high-resolution, real-time infrastructure data. Read more about combining data from all levels of your stack here.
At SignalFx, timeliness and correctness of data is the top priority, and there are a few idiosyncrasies to the API that all New Relic users should be aware of. We’ve done our best to accommodate and we hope these inconsistencies will be resolved soon. Until then, here are some limitations you should know about:
- API rate limit: There is a limit of 3000 metrics total that can be obtained by API calls, which means that not every metric may be pulled into SignalFx. This is not an intentional limit, and New Relic reports that they are working on it.
- API metric name format: In some cases, New Relic metric names include the delimiter “/” as part of the name (for instance: metrics about the endpoint “/*”). In SignalFx, all delimiter characters are treated as delimiters.
- Metric naming: The names of metrics available through the New Relic API may differ from the names as they appear in their user interface. To help guide API users, the New Relic team has written several comprehensive guides about how to calculate common quantities using API results. For example, here are their guides for calculating application throughput and available server memory.
----------------------------------------------------
Thanks!
rt