| Author |
 Topic  |
|
pdrg
Support Moderator
United Kingdom
2897 Posts |
 Posted - 04 April 2007 : 11:55:07
|
TestCount,ExecuteTime,TranType
255,0.17095588235294118,inline
255,0.16495098039215686,proced
Yep, pretty much what I imagined - in this case about 3.5% difference. Don't misunderstand me, I absolutely agree in principle, and personally prefer the elegance of sprocs (hey, a commercial website I wrote back in 2000 for a company has never been data-hacked as it's based on a truly hardcore abstracted stored procedure-based architecture), and there's a good case for sprocs in most cases, but if your site is loaded to the point where 3.5% makes the difference between coping and failing, you need more tin anyway!
|
 |
|
|
Wholsea
Starting Member
21 Posts |
Posted - 04 April 2007 : 12:08:51
|
quote:
Originally posted by pdrg
TestCount,ExecuteTime,TranType
255,0.17095588235294118,inline
255,0.16495098039215686,proced
Yep, pretty much what I imagined - in this case about 3.5% difference. Don't misunderstand me, I absolutely agree in principle, and personally prefer the elegance of sprocs (hey, a commercial website I wrote back in 2000 for a company has never been data-hacked as it's based on a truly hardcore abstracted stored procedure-based architecture), and there's a good case for sprocs in most cases, but if your site is loaded to the point where 3.5% makes the difference between coping and failing, you need more tin anyway!
Keep in mind the small amount of load I was able to put in it, realistically with was maybe around to 10 to 15 concurrent requests hitting the same resource, in a commercial website, that 3% can add up to a hefty chunk.
But yeah, if you are running into that sort of environment, you'd need to load balance your UI side which would be even more of an argument for "host based" data sourcing... |
|
|
|
Topic |
|
|
|
Topic Locked
