Send Custom Metrics to VividCortex

Posted by Baron Schwartz on Aug 20, 2015, 10:40:00 AM

Did you know that you can send custom metrics to VividCortex easily? It's not something we've focused on, but it's been possible to do since the beginning.

One of our agents is a StatsD, which means you can send metrics to it easily with any StatsD client (or none--because it's easy to send metrics to it from a shell script too). The main difference between our agent and a typical StatsD is that you can send us one-second resolution data and it won't immediately fill up disks on a Graphite server somewhere and crash it.

In fact, we send lots of metrics to ourselves about our agents, among other things. It makes it easy to troubleshoot and debug what's happening. You can search and visualize all of the metrics in the metrics dashboard.

Metrics you send to VividCortex should be dot-delimited and the names should only include lowercase letters, numbers, and underscores. They should not include the hostname. The host is included automatically by the aggregator agent. You can send server metrics (if you can find one we don't already capture!), app metrics, business metrics, anything. Internally, for example, our API servers send tons of metrics to the aggregator agent running on the host with them.

We support all the standard metric types, such as counters and timers. Internally we process these as required for our time series database.

Our business model is to be a database monitoring company, not a generic metrics-and-charts-and-graphs company. We assume you want to use us for that first and foremost. That's why we've never really talked about this feature a lot. As a result, we also don't base our pricing on the number of metrics you send us, because unlike a lot of monitoring companies we monitor an absolute truckload of them and haggling over 50 of them at a time is absurd. So, in theory, you're free to send us a bunch of metrics. Really, go ahead. In reality, if you send us enough metrics that you cause a lot of extra load or storage, we'll find out because we are a multi-tenant service and we know which customers are heavier than others. And if it's a problem we'll just get in touch with you and talk about it.

But again, we typically measure many thousands of metrics per second per server already, so go ahead, add some more if they're useful to you.

And let us know how we can help.

Recent Posts

Posts by Topic

see all