A few months back, we updated the account diagnostics page to include a graph showing the number of metrics created over the lifetime of your account. This gives you more visibility over how your account is growing and might prevent excessive use down the line. Recently, we’ve made two additional updates to this page, to give you a more immediate understanding of what’s happening with your metrics. You’ll now see an additional view option, Metrics Treemap, and an additional panel, Busy Metrics Being Ratelimited. Below, we’ll briefly look at how they work and when they can be useful.
Visualise your usage hierarchy with Metric Treemap
You may have noticed that the Metrics section in your account now includes a Metric Treemap. Up to now if you were using, for example, 20,000 out of the available 40,000 metric names for your account, there was no easy way to understand how those 20,000 metrics were distributed — without manually looking through your entire metric tree. To avoid this, we’ve updated your account diagnostics page dashboard to include a Metric Treemap which acts as a visual representation of your tree of metric names and can be viewed one level at a time. It’s hierarchically structured, so the larger the rectangle, the more metrics under that “branch” of the tree. It’s also possible to look inside a particular rectangle by clicking on it, should you want to dig a little deeper.
Additionally, a table at the bottom of the Tree shows an ordered list of the largest subtrees in the current view, limited to the thirty metrics with the highest sub-metrics. It can be used to explore which subtrees are the largest, helping you to trim your metric usage and keep costs down.
Understand what datapoints have been limited with the Busy Metrics Being Ratelimited panel
We’ve written before about how we carefully limit the amount of data we accept from our users. We do this to protect our backend and to make sure one customer’s traffic can’t affect anyone else. As a result, if you send a high volume of datapoints to a single metric in a short period of time, you might be ratelimited. If this happens, it’s useful to know which metrics are responsible. The Busy Metrics panel lets you view these, with a breakdown of the specific metrics we’ve recently ratelimited due to too many datapoints for that exact metric name, should you need it.
In case you missed it…
We launched some other metric related features recently too:
- An open-source system agent that automatically collects system metrics, and offers Graphite-compatible endpoints. Send your Graphite data to localhost:2003 and the agent will take care of authentication, reliable delivery, and compression over a secure transport. Read our recent blog for more information.
- More fine-grained control over automatically expiring unused metrics.
If you’re already a Hosted Graphite user, these new features will be immediately visible in your account diagnostics page. As always, we’d love your feedback—send us a mail to firstname.lastname@example.org or tweet us @HostedGraphite.