Adapt has successfully managed a large number of data centre migrations for our customers onto our managed infrastructure over the last few years.
With every project has come new experience, fresh insight and understanding and each has driven improvements to our overall approaches and methodologies. Following on from my previous blog posts, ‘ Data Centre Migrations – #1 Planning is Key’ and ‘Data Centre Migrations – #2 Secure Your Data’ my final post in the series focuses on the importance of detailed documentation throughout a migration and best practise for physical moves.
For a migration to be successful, there needs to be detailed documentation for all aspects of the infrastructure, both for reference and to plan the new environment. If this exists already, that’s ideal. Most often, there are multiple sources of incomplete/ inaccurate documentation so the task becomes one of consolidating and verifying the data you have into one view and one truth. Try to get as much documentation as possible prepared by the people that know the platform and have been running it over the past few years – they know the environment best.
Physical moves are inherently risky. If there’s any way that can be devised to allow a migration to proceed without physically moving anything in the live environment, it should be explored very thoroughly. In a number of instances, Adapt have moved the entire environment as virtual machines between data centres over the wire. Sometimes there is no realistic alternative to a physical move though, and in these situations there are some points to bear in mind:
Firstly, ensure you are covered by insurance against loss or breakage. Manufacturer warranties normally don’t cover equipment in transit – separate insurance is available that offers this and also following the move provides cover against damage that occurred as a consequence. If the equipment is part of the live environment, break/ fix insurance with very low response times can be arranged.
A special note – many manufacturers insist on being informed of hardware location and won’t provide normal warranty cover if the equipment is not where they think it should be. Some manufacturers insist that equipment is moved by themselves or accredited contractors. It is clearly important to identify and resolve any issues before the move – rectifying afterwards can be time-consuming, expensive and very frustrating.
If you can retire old equipment pre-move, do it. Old kit, running happily, untouched for years in a data centre, can develop component faults that only surface when power is cycled – once you switch off and move the kit, you may find it will never switch on again! Of course you can mitigate against this risk to an extent by power cycle testing in advance of the move. If you do this, make sure you have the ability to fix/ replace any servers that fail.
If you have to move really old kit, you may find that no-one will insure it – failure is too likely. In these circumstances, acquiring second hand spare tin is a good alternative (and with eBay, it can be really cost effective). You can test all this kit without impacting the live environment so you’ll have confidence that if an elderly server dies during the move, there’s a second platform to hand, or a box of working spare parts, whichever is most useful.
If feasible, it is preferable to move entire populated racks. There may be weight or safety considerations that prevent this, but the activity of de-racking and re-racking adds risk and time. A good IT relocation specialist will have professionals that are trained to safely move very heavy equipment.
Following a move, we always suggest running the infrastructure for a period of time in its new location before making platform changes, if this is possible. Any faults that occur as a result of the move are far more easily traced and understood this way.
Whatever your project, good luck with your migration. I hope my blog posts have been helpful to the process and if you have further questions feel free to tweet me below.