Treat your data center migration as if it were the first one ever

December 18, 2019

A blog post by Nicholas Merizzi, Principal, Deloitte Consulting LLP

I’ve spent the last 15+ years helping clients leverage technology to grow their business and the last two helping a major financial services company move its enterprise portfolio to hybrid cloud. In the process my team and I learned a lot of lessons that can apply to a large-scale migration in any industry.

Taking on a large-scale migration can be a daunting (and expensive) task, made more complicated by the fact that an organization’s technology portfolio, internal processes, and end-state needs are unique. These five key lessons may help your organization manage a large-scale data migration effectively and efficiently.

1. The basics are just the beginning.
Before you can set the table, you need a table. A master migration schedule and execution structure form the core around which every other detail will revolve. Like so much else in a process like this, plans like these need broad participation, not top-down compliance. What will be the pace of the migration? What business events may call for a “freeze” to minimize disruption? How will the business and its needs evolve during the project? Who will own each task, and is everyone hearing the same messages?

2. Beware the mega-bundle trap.
Application bundles should stay together through migration, except when they shouldn’t. If a bundle is “too big to fail,” that may be a sign you should break it into components. In one recent experience of ours, that meant handling the mainframe separately. Foundational and shared services such as content repositories and platforms-as-a-service (PaaS) should also stand apart from bundles.

3. Understand the risk, practice for failure.
Every part of every organization has its own risk exposures and tolerances. Decentralizing critical event management to localized IT/business unit partnerships can help align the work with the need. Using a triage approach to staffing tasks and covering resources can provide defense in depth. Knowing critical end-user needs and planning around them—or bridging them with cloud-based solutions—can ward off disruption. And if you approach critical events with migration rehearsals, you can document fallbacks, possibly improve accuracy and timing, and build muscle memory for the real thing.

4. Remember the end user.
The view from the PMO feels all-encompassing. But what about the business unit that depends on a home-grown, macro-enabled spreadsheet? Or the tool that depends on a static IP? Inventorying tools, prioritizing them, and anticipating downstream disruptions can help make sure small-scale issues don’t derail large-scale plans. Because to the people using them, they aren’t small at all.

5. Make the effort to understand your technology estate.
Knowing where you start sounds like basic advice, but if you treat it that way, the execution can suffer. Your starting point is likely complex and studded with islands of tribal knowledge. Many organizations are surprised to learn how much effort it takes to perform the “basic” step of knowing what they have before they change it. But if you don’t address knowledge gaps now, they will persist through migration.

Addressing questions like these helped Deloitte through a recent two-year migration whose scope exceeded 3,500 application instances, 5,000 operating system instances, 1,300 databases, and a thousand mission-critical business tools, with a migration success rate of 99.7 percent and no disruptions. As you approach your own migration, you aren’t the first to confront this challenge. But you’re the first to do it in this place at this time with these resources. Answers are valuable. But questions are priceless.