Modal Title
Machine Learning

Democratizing Data Center Metrics with Intel Snap and Grafana

Sep 21st, 2016 11:23am by
Featued image for: Democratizing Data Center Metrics with Intel Snap and Grafana

In this episode of The New Stack Makers, we take a look at how Grafana and the Intel Snap Framework can be used together to create a killer hardware-based system telemetry service.  Intel Telemetry Developer Associate Matt Brender and Raintank co-founder and CEO Raj Dutt shared their thoughts with TNS founder Alex Williams during the 2016 Intel Developers Forum.

Democratizing Metrics with Intel Snap and Grafana

The discussion can also be heard on YouTube.

Brender introduced the Intel Snap Framework: “We have a standard API  by which you can collect, process, and publish disparate metrics throughout your data center, and do so in a normalized way. That’s interesting, because it’s the first part of a problem of a pipe, getting information out of the data center and feeding it to orchestration engines or to analytics pipe to improve data center efficiency.”

Grafana is an open source data visualization platform that works well with Snap, thanks in part to its pluggable nature. “One of our key kinds of mantras is democratizing metrics, and that means bring data together from all sorts of different places across a company,” Dutt noted. “We’re all about pluggable, composable systems. Plugin architecture really makes it easy for people to contribute, and extend Grafana without having to fork it or go off on their own crazy way.”

As the conversation drew to a close, both Dutt and Brender offered insight as to the state of the IT industry today, highlighting how today’s cloud solutions will continue to change with the advent of open source: “We really believe there’s a lot of roads ahead just in terms of getting aggregates of data. Empower the human being with the right data, the right context to make that decision,” said Dutt.

“If you gather the information and start visualizing it, our pattern mapping brains will do amazing things when they start synthesis up and down the stack,” Brender followed up. “That said, if you feed that right into a system that says if you get 10 hits on this particular problem, go reboot the service, that’s terrifying. You’re asking for rolling outages constantly in the data center.”


Intel is a sponsor of The New Stack.

Group Created with Sketch.
THE NEW STACK UPDATE A newsletter digest of the week’s most important stories & analyses.