Web Operations

Shared, Open Infrastructure

In most situations where operations and engineering have been split into different groups, you'll find that the support infrastructure has also been divided in two. Common examples include:

• Developers running their own dev servers and having no access at all to production hardware

• Engineering using a bug tracker and operations running a separate ticket tracker, with no links between the two

• Engineering and operations running independent metrics gathering and dashboards

• Engineering not being invited to the IRC channel used by ops

• Ops having no access or insight into the source code repository used by engineering

• Engineering having no access to the configuration management systems used by ops

In most cases, this happens for pragmatic reasons: members of one team try out a new system, find it makes their life easier, and never get around to telling the other groups about it. But sometimes the situation arises because one group doesn't trust the other group to not get in the way. Either way, it's a mistake. Having a shared infrastructure is one of the easiest ways to help your teams cooperate with each other. That doesn't mean all teams need to use the same tools all the time; you might find that engineering prefers IM-based chat while ops prefers to use IRC. It also doesn't mean you can't have different levels of access to tools where needed. But you should make sure everyone knows where the other team's tools are and that they have at least a read-only view of ...