Show 246 – Design & Build #4 – Data Center Migration

Ethan
Banks

Greg
Ferro

Listen, Subscribe & Follow:
Apple Podcasts Spotify Overcast Pocket Casts RSS

Moving a data center from one place to another is one of the most complex challenges a network engineer is likely to face. In this Packet Pushers Weekly Show, Ethan Banks is joined by Chris Church and Richard Alexander to discuss just such a project from the viewpoint of having done one or more data center moves in our careers. We go back and forth on approaches to DC moves, reminiscence about the things we did right (and wrong), all while outlining an approach that you yourself can use as a template to move your own data center.

The outline we reference in the show is found below, and is largely the work of Chris. We hope the show is helpful to you as you do more designing and building in your day to day work.

Show Outline

  1. Intro
  2. 3 Approaches to the migration
    1. “U-Haul” Migration aka “Lift & Shift”
      1. Shut it down, move it, bring it back online
      2. Physical moves often take longer than estimated
      3. Labor intensive
      4. Proper insurance coverage
      5. Physical security of the equipment during move
      6. Data protection and integrity
    2. New build-out
      1. Build new and gradually migrate services
      2. Lower risk, higher cost
      3. Maintaining a presence in two data centers can be costly and complex
    3. Cloud/Hybrid
      1. Migration of a subset of services/applications to the cloud
      2. “On-demand” migration
      3. Some services remain on-prem or in colo
      4. Lower capital outlay
      5. Potential for added complexity
    4. Combination of “a”, “b” or “c”
  3. Planning
    1. Facilities
      1. HVAC and power
      2. Connectivity (availability, diversity, resiliency)
      3. Present and future space requirements
      4. Access controls and security
    2. Identify and inventory applications and services.
    3. Identify dependencies
    4. Identify SME’s and stakeholders.
    5. Communication with customers and team members.
      1. Over-communicate.
      2. OOB communcation method for meatspace (don’t rely on the gear your moving for UC.)
      3. Do you have mobile phone (cell) coverage?
    6. Coordination with vendors and availability of vendor support.
    7. Timing and projected duration of service outages.
    8. Connectivity at the new site. Hot cuts vs. new circuits.
    9. External DNS changes.
      1. TTL
      2. Access to make changes
    10. Equipment delivery to new facility.
    11. Physical access for required team members.
    12. Details  (cables, cords, rack hardware, PDU’s, KVM, crash cart, flashlights, etc)
    13. Validation plan.
  4. Preparation
    1. Assemble the team.
    2. Outline the plan and distribute to team members.
    3. Peer review
    4. Develop a checklist.
    5. Gather the required tools.
    6. Communicate!
  5. Execution
    1. Final checks.
    2. Ready, set, GO!
    3. Periodic notification to stakeholders.
    4. Validation.
  6. Lessons learned — post-mortem discussion

About Christopher Church:

Church has worked in the technology industry since the early 1990s, primarily with SME customers. He’s currently an Enterprise Architect for a managed services provider based in Little Rock, AR. He provides design, implementation, and management services for customers globally.

About Richard Alexander:

Alexander is an ex-financial services techie who survived a major core network overhaul and DC migration. Following that, he decided to move into technical pre-sales for the UK VAR arena, and was involved in countless migrations, ranging from moving a single service to the cloud all the way to global DC migrations for international trading platforms. Currently Chief Technologist in Networking, Security and Mobility for Logicalis UK, Alexander is responsible for the Business Unit technology blueprints and technical delivery capability.

Share this episode

Have feedback for the hosts?

We want your follow-up.

Send us follow-up! 😎

Join Our Slack

Chat all things networking, cloud and security in the Packet Pushers Slack community. It's free and open to everyone.

JOIN 💬

Leave a Comment

Comments: 8

  1. Damian on

    one of the best miniseries yet – actionable content, tips from true experience, summary of methodology, and war stories to laugh out loud and feel like we were there with you.

    Reply
  2. Farinha on

    Agree… one of the best Packet Pushers shows I’ve listened to lately. No SDN references either, so double victory 🙂

    Thanks guys for the design and build miniseries and specially for this specific episode! Made me want to listen to the show again to catch all details.

    Reply
    • Ethan Banks on

      Mike, been a while since we recorded this, but I assume it’s the doc we used as our notes to record this show. That Google Doc became the show notes you see above, slightly cleaned up.

      If there was some other doc I was referring to, refresh my memory, and I’ll see what I can find for you.

      Reply
  3. Shivam on

    Lately listening to all of your podcast and have been having amazing time. And this one was another good one. Feels like I was there ?. Thanks for these podcast

    Can we have a broadcast about DR technologies and how networkers have manage that.

    Reply
  4. Allen Garvin on

    This was a fabulous show. I wish I could have listened this way back multiple times in the past.

    One odd anecdote, of one DC migration with a problem never addressed here: We were going from a nearly 10-year old DC that had way outgrown itself, running against power limitations mainly, and for the most part, we got to buy new equipment for the new install, so we migrated most of our storage in advance, and sync it up prior, and we consolidated racks in a 3-1 ratio. And this was Texas in the fall. It had been warm weather all the way until December, when we did the move. The new location was completely new and empty.

    We all showed up in t-shirts and short sleeves with the weather being about 80F, loaded everything up, moved it. Our old data center was hot even in the cold aisles. We got to the new one just as a cold front blew in, dropping the temperature over 50 degrees in a 12 hour window. Plus, the old center was in a well-insulated building, and the new one was stand-alone with no insulation to speak of. About 6 hours in, we were all shivering, just about shutting down from the cold temps. One coworker had a blanket in his car that we shared. Only when someone got sent to a 24-hour target to buy cheap blankets and jackets could we really proceed in comfort.

    That is the strangest issue by far I’ve encountered in any migration. I have a great pic of one of our storage guys, a big 6’5″ guy who played football in college, huddled on the floor under a tiny blanket typing away on a laptop.

    Reply