RE: [sqr-users] (no subject)

["Tritt, Merrill" <MerrillTritt@spherion.com>]

Someone awhile back ago posted some benchmarks between LET and 
MOVE/ADD/SUBTRACT/MULTIPLY/DIVIDE.  Seeing how you have a large volume of rows 
and depending upon what you're doing within your programs, converting some 
LET's to MOVE's will probably speed it up somewhat.

Merrill

===========================
Values adjusted 0.182940 seconds for looping (100,000) overhead.
 
1.337228 seconds to do LET commands.
0.927615 seconds to do MOVE (STRING) commands.
 ===== 
0.669077 seconds to do LET commands.
0.475877 seconds to do MOVE (FLOAT) commands.
 ===== 
1.218661 seconds to do LET commands.
0.392715 seconds to do ADD commands.
 ===== 
1.213833 seconds to do LET commands.
0.403296 seconds to do SUBTRACT commands.
 ===== 
1.156984 seconds to do LET commands.
0.470526 seconds to do MULTIPLY commands.
 ===== 
1.180882 seconds to do LET commands.
0.595128 seconds to do DIVIDE commands.



-----Original Message-----
From: Alexander, Steve [mailto:Steve.Alexander@ci.sj.ca.us]
Sent: Thursday, September 18, 2003 7:29 PM
To: 'sqr-users@sqrug.org'
Subject: RE: [sqr-users] (no subject)


Are you running the conversion once or several times?  Even once is a long
time (1200 minutes or 20 hours), but you may be able to overlap the
programs.  The work to tune the SQRs may take longer.  But if it's several
times, tuning may be worthwhile.

I'm not sure of this advice, so I'd appreciate any corrections or
amplifications.  One option is to run the SQRs from the database server,
with the data files on the database server.  That could reduce the run time
by reducing network usage.  Another option, as James writes, is to remove
the indexes from the tables to which you're inserting.  Indexes help when
selecting or updating data, hurt when inserting data.  A third option is to
get your DBA's advice about commits.  Will your database load faster with
more frequent commits, or less frequent?  You may have to experiment to find
the best approach.

Some conversion programs insert a row and then update it to add data that
wasn't previously available.  Try to have all the data before inserting.
Some conversion programs translate field values ("A" in one system
corresponds to "B" in another).  If you use a database table for that, try
to use a load-lookup to bring it into memory.

-----Original Message-----
From: James Womeldorf [mailto:jwomeldo@fastenal.com]
Sent: Thursday, September 18, 2003 4:13 PM
To: 'sqr-users@sqrug.org'
Subject: RE: [sqr-users] (no subject)


Hi Rajesh,
One thing we sometimes do to speed this type of process up is to drop the
indexes on the target tables, then insert the data, and finally rebuild the
indexes.  Of course you have to be careful not to violate any unique
constraints on your tables since the indexes are not there to assist you
while the insertions are made.
Jim

-----Original Message-----
From: RAJESH SAWANT [mailto:rajsaw@hotmail.com]
Sent: Thursday, September 18, 2003 5:29 PM
To: sqr-users@sqrug.org
Subject: [sqr-users] (no subject)



Hi all SQR Gurus,

Which are the different ways to increase the speed of SQR.

We are using SQR for converting legacy large data ( 200,000 rows in a file )
which is taking around 40 minutes to run single SQR.

Since our conversion process contains 25-30 sqrs the total time for running 
all the
sqrs will take a lot of time.

What can be cone at the  database side or by tuning the sqr to reduce the 
time.

Please help me in this.

Thanks,
Rajesh

_________________________________________________________________
A chance to meet Aishwarya Rai. http://www.myenjoyzone.com/msn/knk.php3 Win 
lucky prizes.

_______________________________________________
sqr-users mailing list
sqr-users@sqrug.org
http://www.sqrug.org/mailman/listinfo/sqr-users

_______________________________________________
sqr-users mailing list
sqr-users@sqrug.org
http://www.sqrug.org/mailman/listinfo/sqr-users

_______________________________________________
sqr-users mailing list
sqr-users@sqrug.org
http://www.sqrug.org/mailman/listinfo/sqr-users



_______________________________________________
sqr-users mailing list
sqr-users@sqrug.org
http://www.sqrug.org/mailman/listinfo/sqr-users