![]() | |
![]() |
| | Thread Tools | Display Modes |
#1
| |||
| |||
|
#2
| |||
| |||
|
|
Hi Allan, Thanks for the response. I'll try your suggestion in several hours when the server is idle. Out of curiosity, why might setting SQL to 1 processor speed it up? There are some packages that have parallel processes with 'max tasks' set to the default of 4. I also tried adjusting the fetch and commit values with no appreciable affect. I checked the memory setting. It's set to dynamically configure and the max is 1906MB out of 2047MB. Is this too high? Thanks, -----Original Message----- A quick one to try is to tell SQL Server to only use 1 processor. Does this help the execution time ? -- Allan Mitchell (Microsoft SQL Server MVP) MCSE,MCDBA www.SQLDTS.com I support PASS - the definitive, global community for SQL Server professionals - http://www.sqlpass.org "Fred Jones" <fredjonze (AT) nospam (DOT) hotmail.com> wrote in message news:012d01c3449b$1dc168a0$a101280a (AT) phx (DOT) gbl... Hi, I'm running SQL2k sp3 on a dual Xeon w/four drives/arrays (SCSI)(OS on C:, Logs on E:, tempdb on F:, Data on G: (500GB)). I have a large DTS package for processing text files for billing and reporting purposes. The input file averages 400,000 rows per day and ends up creating a dozen tables that average 300k rows per file. Various data is exported and zipped to disk with WinRAR. The package is either executed manually on the server or scheduled to run from the server My problem is that the DTS package runs about 20% more slowly on the server above than it does on my desktop workstation (executed locally or as a job), which is a plain old 1.8 P4 with a standard drive. When I check the process both processors are running around 80%. Another example is a single column update on a 1 million row table (again executed on the server) with a single join to an indexed table with 8,000 rows. This update took over 2 1/2 hours. I'm looking for advice and hints about how to start researching theslowness problems. Thanks, . |
#3
| |||
| |||
|
|
-----Original Message----- A quick one to try is to tell SQL Server to only use 1 processor. Does this help the execution time ? -- Allan Mitchell (Microsoft SQL Server MVP) MCSE,MCDBA www.SQLDTS.com I support PASS - the definitive, global community for SQL Server professionals - http://www.sqlpass.org "Fred Jones" <fredjonze (AT) nospam (DOT) hotmail.com> wrote in message news:012d01c3449b$1dc168a0$a101280a (AT) phx (DOT) gbl... Hi, I'm running SQL2k sp3 on a dual Xeon w/four drives/arrays (SCSI)(OS on C:, Logs on E:, tempdb on F:, Data on G: (500GB)). I have a large DTS package for processing text files for billing and reporting purposes. The input file averages 400,000 rows per day and ends up creating a dozen tables that average 300k rows per file. Various data is exported and zipped to disk with WinRAR. The package is either executed manually on the server or scheduled to run from the server My problem is that the DTS package runs about 20% more slowly on the server above than it does on my desktop workstation (executed locally or as a job), which is a plain old 1.8 P4 with a standard drive. When I check the process both processors are running around 80%. Another example is a single column update on a 1 million row table (again executed on the server) with a single join to an indexed table with 8,000 rows. This update took over 2 1/2 hours. I'm looking for advice and hints about how to start researching theslowness problems. Thanks, . |
#4
| |||
| |||
|
|
-----Original Message----- Hi Allan, In SQL properties I enabled processor 1 and disabled processor 2. I also set the dropdown to use only 1 processor. I also tried disabling 1 and enabling 2. In each case the processing times increased from 42 minutes (with 2 processors) to almost 50 minutes with 1 processor. If I manually execute the package on my development workstation (single 1.8 P4) it takes about 36 minutes to run. On the server (with only processor 2 for SQL) I noticed that processor 1 was running about 90% and processor 2 was running about 20%. Is this because of the VBScript in the datapump? If so, any ideas why it runs faster on my desktop that has a cheap drive and only 512 RAM? I guess this doesn't explain the slow SQL updates using query analyzer. Thanks for your time, -----Original Message----- A quick one to try is to tell SQL Server to only use 1 processor. Does this help the execution time ? -- Allan Mitchell (Microsoft SQL Server MVP) MCSE,MCDBA www.SQLDTS.com I support PASS - the definitive, global community for SQL Server professionals - http://www.sqlpass.org "Fred Jones" <fredjonze (AT) nospam (DOT) hotmail.com> wrote in message news:012d01c3449b$1dc168a0$a101280a (AT) phx (DOT) gbl... Hi, I'm running SQL2k sp3 on a dual Xeon w/four drives/arrays (SCSI)(OS on C:, Logs on E:, tempdb on F:, Data on G: (500GB)). I have a large DTS package for processing text files for billing and reporting purposes. The input file averages 400,000 rows per day and ends up creating a dozen tables that average 300k rows per file. Various data is exported and zipped to disk with WinRAR. The package is either executed manually on the server or scheduled to run from the server My problem is that the DTS package runs about 20% more slowly on the server above than it does on my desktop workstation (executed locally or as a job), which is a plain old 1.8 P4 with a standard drive. When I check the process both processors are running around 80%. Another example is a single column update on a 1 million row table (again executed on the server) with a single join to an indexed table with 8,000 rows. This update took over 2 1/2 hours. I'm looking for advice and hints about how to start researching theslowness problems. Thanks, . . |
#5
| |||
| |||
|
|
Hi Allan, I don't know if I'm going crazy. I thought that both processor we're enabled when I unchecked one...but now I'm unable to enable other than processor two. Should I be able to have both processors checked a the same time? Thanks, -----Original Message----- Hi Allan, In SQL properties I enabled processor 1 and disabled processor 2. I also set the dropdown to use only 1 processor. I also tried disabling 1 and enabling 2. In each case the processing times increased from 42 minutes (with 2 processors) to almost 50 minutes with 1 processor. If I manually execute the package on my development workstation (single 1.8 P4) it takes about 36 minutes to run. On the server (with only processor 2 for SQL) I noticed that processor 1 was running about 90% and processor 2 was running about 20%. Is this because of the VBScript in the datapump? If so, any ideas why it runs faster on my desktop that has a cheap drive and only 512 RAM? I guess this doesn't explain the slow SQL updates using query analyzer. Thanks for your time, -----Original Message----- A quick one to try is to tell SQL Server to only use 1 processor. Does this help the execution time ? -- Allan Mitchell (Microsoft SQL Server MVP) MCSE,MCDBA www.SQLDTS.com I support PASS - the definitive, global community for SQL Server professionals - http://www.sqlpass.org "Fred Jones" <fredjonze (AT) nospam (DOT) hotmail.com> wrote in message news:012d01c3449b$1dc168a0$a101280a (AT) phx (DOT) gbl... Hi, I'm running SQL2k sp3 on a dual Xeon w/four drives/arrays (SCSI)(OS on C:, Logs on E:, tempdb on F:, Data on G: (500GB)). I have a large DTS package for processing text files for billing and reporting purposes. The input file averages 400,000 rows per day and ends up creating a dozen tables that average 300k rows per file. Various data is exported and zipped to disk with WinRAR. The package is either executed manually on the server or scheduled to run from the server My problem is that the DTS package runs about 20% more slowly on the server above than it does on my desktop workstation (executed locally or as a job), which is a plain old 1.8 P4 with a standard drive. When I check the process both processors are running around 80%. Another example is a single column update on a 1 million row table (again executed on the server) with a single join to an indexed table with 8,000 rows. This update took over 2 1/2 hours. I'm looking for advice and hints about how to start researching theslowness problems. Thanks, . . |
#6
| |||
| |||
|
|
Hi, I'm running SQL2k sp3 on a dual Xeon w/four drives/arrays (SCSI)(OS on C:, Logs on E:, tempdb on F:, Data on G: (500GB)). I have a large DTS package for processing text files for billing and reporting purposes. The input file averages 400,000 rows per day and ends up creating a dozen tables that average 300k rows per file. Various data is exported and zipped to disk with WinRAR. The package is either executed manually on the server or scheduled to run from the server My problem is that the DTS package runs about 20% more slowly on the server above than it does on my desktop workstation (executed locally or as a job), which is a plain old 1.8 P4 with a standard drive. When I check the process both processors are running around 80%. Another example is a single column update on a 1 million row table (again executed on the server) with a single join to an indexed table with 8,000 rows. This update took over 2 1/2 hours. I'm looking for advice and hints about how to start researching theslowness problems. Thanks, |
![]() |
| Thread Tools | |
| Display Modes | |
| |