Posted by Joe Mellor on 05-May-2014 06:00:00

Data centre redundancy – it’s just like being in the pub

Data centre redundancy – it’s just like being in the pub

You’re in the pub. Well, if you’re reading this, I’m guessing you’re not in the pub. But let’s pretend you are. You’re in the pub and you’re buying yourself and nine friends a round of drinks (I know that’s difficult to imagine, but go with me here). You order ten pints. But then you worry; what happens if on the way back to the table, you drop the drinks? Ten pints is a lot to carry, and it’s very busy in here. You weigh up your options:

You could order another pint and leave it with the barman, just in case. That way, if you were to drop one of your pints, you’d have the replacement ready and waiting immediately. If you order this option, you’ve ordered what you need, plus one more just in case. This is known as n+1.

Alternatively, you could order another ten pints on top and leave them with the barman, just in case. Chances are, if you’re going to drop one pint, you could well end up dropping all ten. If you drop more than one pint, your +1 backup isn’t going to be enough. Therefore you double your order to 20 pints to ensure everyone is covered. This is known as n+n – effectively what you need, and what you need again.

Alternatively, as well as the ten you need you could order another 30 pints on top and leave them with the barman, just in case. This might sound like a large amount but your thought process goes: I don’t want anyone to be mad at me, and I could drop anyone’s drink. Therefore I better order everyone’s pint twice. But what if something happens to the back up drinks? If I can drop them once, I can certainly drop them all again. I better repeat the whole process just in case. This is known as 2(n+n) – what you need and what you need again, twice over!

This might sound like a preposterous (and expensive) thought, but when it comes to power resilience in our data centre, this is the length which our UK data centre has gone to, to ensure 100% uptime is always maintained. Our one UPS is backed up by another, completely separate UPS with an automatic failover [n+n]. However this n+n UPS redundant infrastructure is doubled by a completely identical mirrored infrastructure of another two UPS to create 2(n+n). When using this configuration, to lose UPS backed power, a client would have to have 4 UPS failures simultaneously – and when each UPS offers 99.999% uptime, the chance of this is incredibly unlikely.

Download our guide free guide to choosing a data centre it might not be as thirst-quenching as a pint, but we're sure it'll help you in picking the best, and most resilient, colocation provider for your business. 

FREE download: The service provider's guide to choosing a data centre >

FREE whitepaper: The service provider's guide to choosing a data centre

Topics: redundancy, resilience

Written by Joe Mellor

Comments