[Gllug] Excuses for slow server response times

[Doug Winter]

Simon Perry wrote:
> Even if they employ database caching some queries must have been 
> duplicated over that time. Yet no call was quicker than 5.5 seconds with 
> the slowest being a massive 68 seconds.

For a lot of event-driven architectures pre-emptive caching is very 
difficult or impossible - I have encountered this very problem myself.

However, the numbers you quote make this seem unlikely.  If you were the 
only person using the service then they'd look realistic - 5.5 seconds 
for a complex query isn't insane, and if you were the only person 
executing it then that would seem right.

However, I'd expect that mostly you'd get < 1s responses with the odd 
huge one sprinkled in, as you were unlucky enough to hit the service 
when the cache was invalid.  From what you say, the idea that you are 
the only person using it is very unlikely.  Since your minimum response 
time is 5.5s I reckon there's probably something else going on as well.

This doesn't mean they are fibbing of course - a lot of places employ 
nobody with clue in scaling and operational stuff like that, since it's 
rather hard to learn (you have to pick it up on the job).  Be gentle 
with them :)

Cheers,

doug.

-- 
6973E2CF: 2C95 66AD 1596 37D2 41FC 609F 76C0 A4EC 6973 E2CF
http://adju.st/
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug