This blog is a behind the scenes look at what we’re currently building at Dataloop. The new self-service account model!
- Custom monitoring is too complex and time consuming to setup
- Open source tools don’t scale when you combine metrics from ops, dev and the business into a single system
- Ops and Devs end up using different tools so you end up with data silos and no overall visibility
We’ve already solved the scaling issues and built a platform so that Ops and Devs can live happily within a single tool.
What we want to do next is solve the fundamental problem of setup, adoption and engagement. How do we reduce the burden of setup on Mr Monitoring (you know you have this person) and how do you get everyone engaged with the real time data being produced by your systems in order to improve the quality of your online service.
The answer is simple: Distribute the setup of your monitoring out to your teams. Devolve responsibility down to each product team but let your central Ops / Platforms / Systems / WebOps / DevOps / [insert name du jour] team have oversight and control.
What most organisations have been asking for in a round-about way is for a Github style account model. Users should be able to move freely between projects, or even organisations, and they should be able to contribute to the setup of monitoring for whatever they have been given access too. While also potentially being able to see but not change other areas. For the central platforms team they want to be members of the ‘owners’ group so they can dip in and provide some oversight and even pass down the costs of monitoring to their teams in larger organisations.
What you’ll get in future version of Dataloop is an initial screen that looks very similar to the screenshot below.
For Company A you’ll get an overview of the total number of agents, tags, plugins, dashboards, rules and members. You’ll be able to see all of your monitoring ‘repos’ and an event feed of what your colleagues have been doing in Dataloop for the areas you have visibility over.
We think of repos as a bit like Git repos as they encapsulate monitoring configuration: plugins, dashboards and alert rules. In the world of micro services these might match your team structures allowing each team access to setup their own stuff. Within each repo you still have tags that can be used to split out your server environments and roles.
There will be an easy to use role based access control system that your central platforms team can administer to provide access to each individual depending upon what they need to do in Dataloop.
At the repository level we’ve rewritten our UI from scratch to support proper multi user participation. Any action performed in the web UI is immediately updated on everyone else’s screen in real-time. You can write plugins in the browser, test them by running them on servers, deploy and then create dashboards and alerts collaboratively, in real-time, without conflicts.
We’re pretty excited about these next set of features. If you want to get onto the beta ping us an email at firstname.lastname@example.org and we’d love to get feedback!