January 9, 2010 at 2:58 am
I saw huge performance improvement after implementing 'quirky method'. Yes, The execution time was reduced from 45 minutes to 30 seconds
I noticed one thing in the execution time.
Before Fine Tuning:
Running the same query from onsite(US) with same parameter, took 5 minutes only.
Running the same query from offshore with same parameter took 45 minutes.
How this much of difference? I am wondering the time difference. is there any problem with network setup?
After Fine Tuning:
Running the same query from onsite(US) with same parameter, took 30 seconds
Running the same query from offshore with same parameter took 27 seconds.
karthik
January 9, 2010 at 7:29 am
I can't be sure because I was actually there to see everything happen, but it probably has to do with the number of reads before and after tuning. Queries that need to "talk" alot will obviously be much slower over a long haul connection than a much shorter or direct connection.
--Jeff Moden
Change is inevitable... Change for the better is not.
January 27, 2010 at 10:49 pm
Jeff,
Code has been moved to PROD. Execution time is 26 seconds.
Dynamic Sql Part : 13 Seconds
Quirky Update(# table) : 6 Seconds
Real Table Update (by using the above #table) : 7 seconds
But the end users expects to finish the procedure within 5-10 seconds.
I am looking the dynamic sql part now.
is there any way to reduce the execution time of 'quirky update' ?
karthik
January 27, 2010 at 10:51 pm
Network team is looking Latency issue.
karthik
January 28, 2010 at 6:54 pm
karthikeyan-444867 (1/27/2010)
Jeff,Code has been moved to PROD. Execution time is 26 seconds.
Dynamic Sql Part : 13 Seconds
Quirky Update(# table) : 6 Seconds
Real Table Update (by using the above #table) : 7 seconds
But the end users expects to finish the procedure within 5-10 seconds.
I am looking the dynamic sql part now.
is there any way to reduce the execution time of 'quirky update' ?
Heh... now wait just a minute... In all the code blocks contained on this thread, I don't see any dynamic SQL directed at the resolution of this problem and the "Quirky" update is the fastest of the 3 things above... I'd concentrate on the dynamic SQL part first, whatever it is.
Also, how many rows are we talking about the "Quirky" update doing?
--Jeff Moden
Change is inevitable... Change for the better is not.
February 2, 2010 at 10:09 am
Also, how many rows are we talking about the "Quirky" update doing?
300000 rows(nearly).
karthik
February 2, 2010 at 5:56 pm
In that case, I guess I'd need to see the code, the tables, and the indexes for all 3 sections. I have a 7 year old single 1.8 GHz box with only a Gig of ram and the quirky update will do a million rows in the same time you say it takes to only do 300,000.
--Jeff Moden
Change is inevitable... Change for the better is not.
February 11, 2010 at 7:18 am
Jeff,
For time being They agreed the current execution time. But i need to look into this for next month release.
Meanwhile i have been asked to fine tune one more procedure. I am looking into it. I will get back to you shortly...Thanks a lot for your help 🙂
karthik
Viewing 8 posts - 16 through 22 (of 22 total)
You must be logged in to reply to this topic. Login to reply