I must have been asleep when Microsoft published the Exchange 2010 datacenter switchover tool last October as I have utterly failed to acknowledge its existence until now.
[Note (May 2013): Microsoft has replaced the PPT-based version of the tool with a more elegant web-based version. However, the basic concepts still hold.]
Some have noted that this is not really a tool at all, largely because it has not been created by coding the steps necessary to switchover a Database Availability Group (DAG) across datacenters in some form of computer code and presenting the resulting knowledge as an executable program. The assertion is true, but only if you regard a tool as something that absolutely has to be an executable. I guess this is true, if you assume that tool in this context means a computer utility, but it’s not in the wider context if you regard a tool as something that helps you to do a job more effectively, which is what this tool does.
Of course, you might regard PowerPoint as a language. Certainly, there are people who can make PowerPoint do things that I would never dream of, such as the PowerPoint MVPs. So perhaps the datacenter switchover tool, which is provided in the form of a PowerPoint presentation, really is a tool after all.
Now that we have battered the semantics of the situation to death, the tool is the result of some pretty impressive work by Tim McMichael, someone who is well known as a true expert in the arcane art of the lesser-known corners of DAGs. Tim has presented at many conferences and his sessions are always content-rich and full of practical knowledge, something that comes no doubt from his day job as Microsoft’s go-to guy for DAG problems. Indeed, as he’s remarked from time to time, if you have a DAG problem and Microsoft support directs you to Tim, you are so far in the brown smelly stuff that you won’t notice it flying around your head.
Grabbing the knowledge that exists in someone’s head and encapsulating it for reuse is the domain of “knowledge engineering”, an activity that was all the range in the mid-1980s when the artificial intelligence community attempted to work out how they could extract knowledge from humans and make it more generally available through computer programs. I participated in some knowledge engineering projects when I worked at Digital’s European Technical Center at that time. Apart from the ALL-IN-1 Mail Filter, we never achieved any great were achieved, but it was great fun to try to persuade real experts to liberate their knowledge in a way that it could be accessible to others.
Tim’s PowerPoint presentation might seem crude, but it works. Like any recipe, you have to bring your own knowledge and experience to the task in order to be truly successful. Tim lays out the basic framework for making a decision as to how to best perform a datacenter switchover for a DAG and how to switch it back once the underlying problem is solved. Like any program written by someone with no knowledge of your environment, you have to take the recommendations presented by this tool and put it into context. You know the details of your production environment, how the databases are laid out on servers, how many copies exist, what file share witnesses are in place, and how the network functions. You can follow Tim’s directions to the letter and I have no doubt that you’ll be successful, but it’s always best to engage the brain into forward gear when coping with complex issues.
Once they scale up to more than three or four servers, DAGs can be complex beasts. This is especially true when they span multiple Active Directory sites and multiple physical datacenters. This is a really interesting and valuable tool. Anyone dealing with a complex DAG should at least look and learn. You know it makes sense.
Follow Tony @12Knocksinna