If it isn't broken, don't fix it?  Or should you?

So you have a bunch of servers, one doing Exchange, a few doing websites, another doing some super-secret ninja stuff and one hundred desktop computers.  They are all working perfectly - today.  The only issue is you are just about to pass the three year mark of which your warranty will expire on the servers.

What do you do?

Do you replace the servers?  Do you renew the warranty for another year or two?  Or just let it keep going in its current form?  Or are you about to leave the company and don’t care (the next guys problem™)?

Servers are not usually cheap, unless you go for a bunch of white boxes.  Then you probably have other issues, like your supplier not usually having stock of the motherboard for when it decides to blow a capacitor at 2am and you may be without an Exchange server for a few days at least.  Sticking to a nice vendor, with an even better support contract, you know when you do call them and tell them it is broken – they will there within hours with enough bits to rebuild two servers. 

But what does it cost you, if your Exchange server blue screen crashed and you had to run a check on the information store?  Average wage… let’s say is 20 bucks per hour by one hundred users… that is around $4000 bucks for a two hour outage.  Even more so if you count how much business might have been lost via email like if a quote had to be submitted by lunch time.  What does a server cost?  $4000 can get you a half decent server with a bit of memory and fast hard drives.  In business terms – that is nothing.  Triple the money and you will two servers which are twice as good.  What happens if you had to restore the entire server from a two day old backup because the power supply caught fire?

What does it cost you to replace the server early – like 3 months before the warranty had finished?  Time wise, it still be less than a single outage.  Maybe a long weekend to replace the Exchange server as most of that time will be waiting for the users to migrate from one server to another.  Server costs?  They will probably cost more than a single outage but should be less than a few outages.

Maybe it is the perfect time to adjust how you do things?  Maybe it is time to think about having an Exchange cluster?  Even if one of the servers were older and the other newer – they would still be able to pair up to help reduce the impact of an outage.  There will come a time when the older server will need to be replaced still – but this could save you sometime today to be used at a later date.

Or going one step further, maybe chucking a few mega servers in with a dedicated SAN/NAS behind them running some flavour of virtualisation could be put in?  This does cost a lot more than just a few new servers to replace older systems, not to mention even higher risks – like what if you under plan your systems and then you have bad performing system which costs you real money in lost productivity. 

There are risks everywhere.  Even getting a new server to replace an old one – you have the risk of a failed migration or extended outage due to unforseen issues.  What if the original I.T. guy did something really weird or stupid with the original server that you didn’t know about and it caused major issues for the replacement?

Maybe we should change “If it isn't broken, don't fix it” into “If it is getting old, and has a risk to cost the business a lot of money in lost productivity, and the cost to replace it is less than two major outages then we should just do it”?  

Risks are everywhere – it is just a matter of how they are dealt with and how much we can reduce the possible impact to business.

Posted in: [Miscellaneous], [Network Infrastructure / Architecture]