5 Major Challenges in a Data Center Migration (Part II)

Jerry Gentry is the VP, IT Program Management at Nemertes Research

In my last article I talked about the Five best practices in data center migration program. I recommend establishing new environmental needs for power and HVAC, plan the new layout, get all the pre-work of cabling telecom and electrical done, start creating an inventory that will be accurate and complete by the “go live” date and plan out the migration process well in advance. As in any project, there is also the dark side. Today I will cover the five things to watch out for in a data center migration.

Five Things to Watch Out For in a Data Center Migration

First, hidden complexity will hit you. You probably do not know all of the back-end attachments to the primary applications you are going to be moving. There are legacy applications sitting in your current data center that are older than you. It is never too early to start a detailed inventory with your business customer to track everything down and make sure you have an owner. All the information you discover needs to find its way into a CMDB type database—not a spreadsheet on someone’s laptop.

Second, post-migration testing is a challenge. But, since you are talking to your customers to map things out, you have an excellent excuse to start the conversation about how they are going to test the applications before and after the migration. (Why before? Like any doctor, you need a baseline on your patient. You need to know how things actually work, not how folks think they do.) Enlist network staff to time performance end-to-end on a specific set of transactions on key applications. Document those tests, then repeat them after the migration. Nothing stifles whiny end users better than facts.

Third, migration breaks regular work schedules. Start informing your end users and support teams that some of them will be putting in overtime to do the QA needed to support the migration. Let’s face it, you’ll be constrained on when you can move certain applications because of application owner freezes and critical process times. The migration scheduling alone will require months of planning. It is never too early to start, but expect overruns in overtime.

Fourth, application delivery optimization (ADO) is fragile. If you use load balancers or optimizers (two different ADO technologies) you’ll have to peel back the layers of their configurations and understand how you are going to manage the migrations. This may require some additional investment for duplicate hardware you weren’t expecting. Look for changes you can make in the current configurations to build greater modularity. That requires a plan for the migration to be worked out.

Fifth, what’s buried isn’t usually treasure. Somewhere in all those applications and back-end databases you will find some hardwired IP address or domain names. Not only should you start ferreting them out at once, but you can also use this opportunity to position yourself for IPv6 readiness. That means you should have an IPv6 strategy worked out and use it as a reference during the application and network component review.

Well, there you have it. There are many more elements that comprise the data center migration, but if you focus first on the Five Best Practices and Five Things to Watch Out For, you will make great strides to assuring success and happy customers.

Jerry Gentry is the VP, IT Program Management at Nemertes Research