Datacenter and Workload Migrations: Lessons from the Trenches

If you have a data center migration, or consolidation, there are some very specific things you need to do to prepare to make that project a success.
Written by Gery Menegaz, Contributor
If you have a data center migration or consolidation, there are some very specific things you need to do to prepare to make that project a success. And if you are moving from one provider to a new provider you should not start the project until you have met certain planning criteria.

The lessons learned from these sorts of migrations are significant for several reasons, including IT internal and external customer satisfaction, cost, and general success in winning business for integrators.

By their nature, these sorts of projects tend to be especially stressful and team building and sense of humor become critical tools. Engaging the technical teams from all three organizations early on in the process will facilitate knowledge transfer and foster teamwork.

Communication and sponsorship with the customer and 3rd party sub-contractors is critical; communication with all levels of the 3rd party organization become critical to success in these sorts of engagements. The project will be equal parts staffing, management, communication, negotiation, and architectural challenges. Be flexible in action and creative in thinking and be prepared to break with established processes in efforts to move the architecture work forward.

The first hurdle will be to ensure that the existing and new providers, customer, and anyone else who may have a stake in the project are involved and on the same page with regard to the work that is contracted to be undertaken. Any agreements, no matter how trivial, must be documented as memories will fade after a few months.

Contracts are a tedious and take a good deal of effort to draft. If you have a team on the bench waiting for a contract to be drawn up and accepted, expect to have the project run over budget. The rule here is, without a proper agreement in place, the project should not start as this will burn project funding and cause budget issues.

Once the agreement is in place, experience has shown that while we will find day-to-day technical resources more than cooperative, a 3rd party vendor’s management will likely be resistant especially if they are the one’s losing the business. The result of this will be that the project timeline will be pushed out due to escalations on trivial matters, transition activities, and competing priorities.

The processes that one provider is accustomed to employing will likely need to be modified; which may require a modification of the contract in some cases. The process for migration will likely be driven by the project team which is a mistake. Just because you have a crack technology team who can move hundreds of servers over a weekend does not mean that the client or business has the bandwidth to test, validate and accept that volume of change in the environment.

Additional project management overhead, as well as inclusive communication strategies, and creative thinking will be required to succeed. You must anticipate difficulty in simply scheduling meetings with client and providers, architecture and technical SMEs as priorities may not always align.

When the migration project is not a priority for the teams, I have found, inadequate or non-cohesive project management (across any of the 3 players) will result in more delays. Keep in mind that if the goals of the three parties’ is not aligned, then simple tasks like transferring licenses or establishing network connections from the new to existing provider environment will require escalations.

Typically, providers will not allow any 3rd party resources to enter their premises, especially if the provider is hosting multiple clients on share infrastructure. This means that you will not have direct access to management tools, nor servers to collect data impacting solution framing timeline.

Without access to the data center and management tools, you will need to rely on vendor resources to engage to do some of the work; this will result in shifting of work from one in-house or new provider to the existing provider. Shifting resources means a scope change, and may push the project dates to the left.

The very same security concerns that keep you out of their data center will likely also prevent you from running data collection and migration tools. Moving from an automated discovery to a manual discovery can and does add months to an engagement. This is because the data that the customer perceives the current provider to have is often not complete. Additionally, environments are typically more complex than the customer understood. The knock-on effects are:

  • Non-functional requirements may need to be modified
  • Architectural decisions may need to be revisited as new data is provided
  • Technical design may need to be modified as new data is made available
  • More resources required during test phases than expected due to complexity of environment

In the rare occasion where they do allow for automation tools, the existing provider’s management may want to review any customer data that is being provided to a 3rd party. This process involves potentially some portion of data not making it into the hands of the new provider’s resources. What is provided is typically, though not always, is inadequate and dated.

Lastly, the new provider should be prepared to receive only what was requested; therefore, requests must be very specific with regard to what level of detail is expected.

Don’t simply ask for Utilization Data; ask for CPU Threads and Utilization, Memory and Adjusted Memory, Network I/O, Disk I/O, Disk Space Usage, etc. Also, because data collection may need to be revisited, any tools required to do an assessment ought to remain available for some time after data collection is complete, ideally through the sign-off of the technical design.

Editorial standards