Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
mmilo
Starting Member
2 Posts |
 Posted - 2007-05-21 : 07:10:46
|
| I read every single topic about performance and long transactions in sql server without getting any valid solution.I have a .NET application running a long transaction writing thousands record "randomly" in up to 10 tables.I monitor the process with a progress bar and a counter of record wrote and notice a repetable and evident performance loss during the process. Suppose I manage up to 100 record/second at start up, after 30000 records the performance fall down to 3 record/second .Anyone can help me?thanks |
|
|
Kristen
Test
22859 Posts |
Posted - 2007-05-21 : 08:18:07
|
| Rebuild the indexes?Kristen |
 |
|
|
harsh_athalye
Master Smack Fu Yak Hacker
5581 Posts |
Posted - 2007-05-21 : 08:19:56
|
| How rebuilding indexes going to improve write performance, Kristen?Harsh AthalyeIndia."The IMPOSSIBLE is often UNTRIED" |
|
|
|
Kristen
Test
22859 Posts |
Posted - 2007-05-21 : 08:24:28
|
| Loads of index page extensions after inserting lots of records, which equates to either very badly fragmented, or largely "sequential search" perhaps?Or each insert having to force a page-split because pages all full (in which case need to rebuild with sufficiently "loose" FillFactor)Is that a reasonable assumption, or am I barking up the wrong tree?Kristen |
|
|
|
mmilo
Starting Member
2 Posts |
Posted - 2007-05-21 : 09:36:19
|
| Kristen, you are right.I wonder why if I commit and restart the transaction the performance initially boost (max speed) with the same side effect of losing performance running.Anyway I'm testing index rebuild. thanks |
|
|
|
|
|
|
Long transactions slower and slower