Rollback vs. Roll Forward The Advocates at Redgate Software had an interesting discussion about deployments in databases and how you go forward or back from the point at which you discover a problem. You can watch the episode, but a few things occurred to me while we were having our discussion. The first thing is we all agree data makes things hard. A database is a stateful object, and dealing with stateful objects is hard. That is one of the things I’ve internalized the last few years that has tremendously changed how I work with Redgate customers. The more I consider state, the more I am able to work with the challenges that databases bring. The second interesting thing from the episode for me was that each of us had a tendency for what to do. Do you tend to aim to get to a previous state or move to a new state? Each of us had a preference. I think lots of us aim for a new state, mostly because we’re optimistic about our ability to “fix” the broken thing. I also don’t think this is a DBA, engineer, or technology thing. I find lots of people in the real world making mistakes and thinking they can do a new thing to fix the situation. Mechanics, lawyers, doctors, plumbers, they all think they can roll forward to a new state. What’s occurred to me is that the people I know who are very much in demand for their skills and expertise are often the ones with a tendency to roll back to the previous state. They acknowledge the mistake and undo it. I’ve seen people in construction and other professionals also try to go back to the previous state when they realize they are in a situation where their fix didn't work, and they abandon that plan. In the DBA world, we might prefer this even when our deployment caused data changes. We often will save data, roll back, and then decide what to do. This often means reconciling data, which isn't a fun task, but a necessary one. Watch the episode and decide what your tendency is for adjusting for changes. Do you think about patching the issue and rolling forward? Or do you want to roll back to a known state and re-test our changes in a lower environment. Steve Jones - SSC Editor Join the debate, and respond to today's editorial on the forums | 
