Re: You couldn't make this sh** up
> Go on then, how could it have been predicted?
Well firstly, as has already been mentioned, things are known to get busy leading up to a deadline. It's not like this has never been observed before :
http://www.theregister.co.uk/2014/10/01/dvla_website_outage/
And the one I can't find offhand about HMRC allowing extra time for self assessments after their site failed to handle deadline demand.
Some numbers are (or should be if the systems analyst wasn't a complete numpty).
Total number of people in the UK eligible to vote. Exact numbers probably aren't, but I bet there are people with a darned good idea.
Total number of people actually registered. If this isn't known then we might as well pack up and go home.
Subtract one from 'tother and you get a fairly good estimate of how many aren't registered.
As above, you should be able to watch trends, and apply knowledge from previous events, to get an idea of how application rate is likely to scale up as the deadline looms.
I think the only unknown is the number of applications it'll be expected to handle from people already registered - but then, why isn't it also handling that part of de-duplication ?
This is precisely the sort of thing "cloud" is supposed to handle - with the ability to scale up a (properly designed) system very quickly - and shut it down again when the peak is over. That suggests that either the people responsible are clueless f***wits, or the people responsible for them and their budget* are clueless f***wits.
* As has been pointed out before, it's possible that external constraints precluded a scalable system in favour of fixed costs. Whoever was responsible for that should be hung out in public.
In short, no - such peaks are neither unprecedented, nor are they unpredictable. A competent design would have coped with it.