We may be able to conquer the moon, but it is proving harder to beat downtime
As you are no doubt aware by now, IT Week has changed, and looking at the new
full-length picture of myself, it looks like I have as well. But before you
start racking your brain trying to think of
The Sweeney character I’m trying to
model myself on, I’ve got just two words to say: “Shaaat it!”
When I realised my column was going to grace the first issue of the new-look IT Week, I knew I had to come up with a topic that would make readers sit up and take notice. My list of candidates included holographic videoconferencing, wireless power transmission and lunar datacentres.
OK, that last one may be a bit far-fetched, but I firmly believe that at some point in the future there is going to be a permanently habited base on the moon, with its own dedicated network manager. And I guarantee that he or she will still have to deal with most of the standard causes of downtime we get down here on earth. Maybe they won’t have to worry about a builder turning up at reception to apologise for putting his spade through the network connection, but I suspect they’ll be familiar with all the other things that cause network grief.
Downtime comes in two flavours – planned and unplanned – and it’s normally the latter that leaves a nasty taste in the mouth. However, it wasn’t always so, especially in the days when very few firms had their telephony going over the same network as their email.
I can remember, back in the late 1980s when I worked in academia, trying to harness the power of Fortran 77 to process some data. I had a VT100 terminal, connected to one of Digital Equipment’s venerable Vax systems. When the Vax crashed, us eggheads could go down the pub for a couple of hours because all our applications, including our rudimentary email system, were run on that one system. These crashes could be irksome but unlike today, they did not mean the work of the college had to grind to a halt, because most departments were not reliant on IT.
Unplanned downtime today is a much more serious problem. My old colleagues and I could survive quite happily without email for a few hours, but I can’t imagine many businesses today being able to do that. As more firms start to rely on their networks for voice communications as well, unplanned downtime will be as welcome as a tornado at a camp site.
Today, single points of failure in hardware, software and networks are being weeded out by firms in a bid to achieve the highest levels of availability possible. For larger firms with the budget to throw at such things, clustered databases and hardware that automatically fails over to a remote site should be able to ensure application availability to at least three nines. However, for smaller firms, it looks like virtualisation could be a bit of a business-saver. A server OS failure should trigger a software failover, and another virtual server OS popping up to take on the IT duties.
But what if you’re a small business and the server hardware fails? Well, the duration of the system interruption will probably depend on whether your IT crew are up to the task of fixing the fault – something that is sadly not always the case.
The only way for smaller businesses to plan for outages like this in the long run is, I suspect, to become bigger businesses with the financial muscle to afford some hardware redundancy. Alternatively, they could turn to one of the many managed service providers that claim to offer four-nines service levels.
