Ummm……don’t want to be too dispiriting about this article but it is actually quite easy to achieve six nines of server availability, let alone the 5 nines that the author is aspiring to. It is not complex and does not require “massive hardware and systems”.

Simply substitute your industry standard server with an industry standard server. Yes, read that again! By simply replacing your server with the same which happens to be designed as a fault tolerant system, you can achieve around 30 seconds downtime a year – a significant improvement over the 5 minutes.

Now, does this cost more? Well of course it does as there is more “stuff” in the server to make sure it can withstand failures – but not as much as you would imagine – we are not talking about mainframe pricing or silly entry level “non-stop” pricing here.

And how about avoiding that unplanned downtime? Well, that’s something that virtualisation technology brings to the table. So imagine the situation – something goes wrong in your industry standard server – bang – all the app’s fail and have to be restarted. This equals downtime. With fault tolerance, yes, the server still goes “wrong” but it carries on processing.

Live migrate your applications to a spare box (even this is not strictly required as all components are hot-replace – however, for system updates/reboots then yes), do your updates, reboot, bring back online, migrate applications back over. No downtime at all for either component failures or system updates.

Related Articles