Jason Dixon - Monitoring with Graphite

Three years ago I was writing about how #monitoringsucks and we need to reconsider our approach to monitoring architectures. Traditional monitoring systems were monolithic and unwieldy. Nagios already had a largely negative connotation but there were few alternatives on the horizon.

Either by luck or instinct, I correctly predicted that the monitoring ecosystem would evolve towards a composable model of well-defined services and compatible interfaces. Look at the Open Source monitoring community and this is precisely what youll find today.

In particular, the emergence of NoSQL databases has resulted in enormous momentum and competition among the various time-series databases. Users and businesses have embraced Etsys Measure Anything, Measure Everything mantra, which has driven the need for larger and faster data stores. A hugely successful conference series arose out of the desire for more collaboration and discussion around our monitoring toolchain. People are more open and transparent about their processes and tooling and these shared experiences benefit everyone.

In other words, software moves fast and Graphite is no exception. Fortunately, Graphite has reached such a high level of adoption among users and businesses that youll be hard-pressed to find tools and services that dont integrate with it directly through its robust APIs or among any number of third-party bridge utilities. In many regards, Graphite has become a time-series specification that other software projects leverage for themselves.