An article by Cliff Moon of Boundary on GiagaOM triggered a small avalanche of -sometimes hostile- responses as well as a post by Brandon Burton. We get a lot of our motivation and ideas from the #monitoringsucks community and I couldn’t agree more with Brandon when he writes:
This perfectly describes our application environment and no doubt that of many other SaaS companies, and it’s why Metrics was not built to be a “soup to nuts” monitoring solution, but as a platform to which developers, DevOps and operations teams can connect the monitoring tools of their choice.
It’s our belief that decoupling the collection of time series data from its management (i.e. the storage, visualization, analysis and automation functions) is essential for realizing the “monitoring is a toolbox” future, as it makes it possible to put together a monitoring infrastructure that is adapted to the specific and changing needs of a particular environment, while providing a unified solution for storage, analysis, and automation.
In a tool box world companies can choose whatever time series data management solution they prefer, be it a SaaS like Metrics, or a self-hosted solution based on tools like Graphite and Nagios.
It’s our vision that the future of monitoring will bring a large universe of community provided and commercial monitoring tools; a choice of tools and services for managing and analyzing monitoring data, and a uniform way of connecting monitoring tools to any of these data management and analysis platforms. This will enable DevOps and operations teams to select, and in some cases build, the data collection tools that are best suited for their environment and couple these to the storage and analysis tools that best fit their needs.