2 or 3 things I learned building a monitoring service

Abstract:

In 2010 3 of us started to build a product for the devops crowd, predicated on a few napkin drawings and 2 basic ideas: (a) software engineers who write software for scale, live and die by metrics and (b) these metrics come from a zillion different tools that create silos. We chose a memorable name and a cute logo and off to the races.

Fast-forward 3 years, the 3 have become 15, the product has evolved into a monitoring service and the company is gaining new customers every week without any marketing.

In these short 3 years we have had to:

• find our customers (you, here) and find our focus,
• scale from a single postgres instance to 5 different datastores all working in concert,
• swap out parts of the backend as they become bottleneck
• build X-ops into the core of the company (where X in [dev, tech, customer]),
• discover a whole set of new metrics to obsess about,
• learn something new every day, really.

I would like to share with you some of the lessons I have learned (sometimes the hard way) in building this monitoring service because I wish someone had done the same when I started.

Speaker:

Alexis Le-Quoc

Co-founder and CTO of Datadog, a monitoring service (that does not suck ;).