Build vs. buy#build-vs-buy

Building your own incident response solution can take a lot of effort, including planning, development, documentation, and ongoing maintenance. While many organizations have built their own incident response tooling in house, be aware of the resources and maintenance time you'll need. These resources could be devoted to building products instead. Buying an incident response solution often makes more sense than building one, but it depends on the problem you're trying to solve.

If you need to make a small procedural change (for example, automate a single process) and you have the resources for building your own tool, it might be worth a try (assuming that the in-house tool you build can be maintained and updated without taking precious engineering time away from other high-profit opportunities). However, there is no guarantee that a homegrown solution will be up and running if you have a massive outage.

On the other hand, buying a tool takes away your control of how the tool functions and your ability to build new features into it. However, a vendor building the tool you need has been doing only that--focussing on all of the needs, use cases,e and possible extensions of that tool--so the specific functionality you need at your organization is probably there already. And when you buy, you don’t have to worry about maintenance or scalability. It’s also important to note that tools you can buy are already designed to accommodate DevOps teams who are working on multiple features under time constraints. This often means that buying a tool that ensures reliability, reduces process friction and helps to foster positive behavioral change on your team.

Our take?#our-take

Don’t build anything that isn’t core to your business. It’s more expensive than you’d think! A little back-of-the-envelope exercise: an engineer getting paid $150k, at a minimum, to build this tool would take six months to build it. Maintenance of the tool will take 30% of that engineer's time. That's $75k spent building your tool alone... and a considerable annual cost to maintain it.

And what if your in-house incident management tool goes down? How are you going to run an incident for your incident management tool?