Data Center Migrations – Advice from a Pro For The Team

The cold reality is that Data Center Migrations are hard. Each migration is a unique and complex undertaking where literally one wrong move (pun intended) can leave you months behind and millions in the red. That danger is magnified by the fact that most entities face this task as first-timers. Like open-heart surgery, it’s not something you do every day and certainly can’t afford to screw up. 

Here are some tips from the trenches…

1)     Discovery and analysis data is:

·       Defined as to what is needed (so you know what to collect and when you’re done)

·       Kept in one central place (database or huge spreadsheet)

·       Updated religiously

·       Easily sortable by scenario/perspective (system, move date, owner, disposition, status, etc)

2)     Your migration event is:

·       Planned down to minutes with someone allotted to monitor status the day of

·       You can look at the plan from both a staff and server dimension in case problems arise with either

·       Has periods of slack built in to recover from delays

·       Includes a defined contingency factor & point-of-no return

·       Has steps already defined in the event of retreat

3)     Your move date is:

·       Communicated broadly, repeatedly, and well in advance

·       Includes an agreed upon fallback date in case a plan B is needed
 

4)     Backups are:

·       Required – Lost data is lost, disks fail more than anything else in transport

·       Verified – Test to make sure A) the backups are good and b) you can restore at either location

·       Duplicated – One set can be lost/damaged/corrupted

·       Planned For – They often take quite a bit of time, even just for deltas

·       Shipped securely – Each set is shipped separately, most companies will need to encrypt as well

·       Often overlooked once migrated – Make sure you remember to test backup service at the new facility once you’re moved in too

5)     Most common 11^th hour problems are:

·       Network design mistakes/gaps (e.g. firewall rules, re-IP)

·       Insufficient cabling (missing drops, short the necessary cables)

·       Power (missing whips, improper plug types, proper redundant power)

·       People (tired, unreachable)

·       Server Access (account privileges, server management, access behind firewalls)

·       Latency “Surprises” (slow apps… e.g. needing Citrix)

·       Sudden critical processing windows (end of month/quarter/year processing, customer events, audits)

·       Procurement Delays