Hi all
I want to ask you something about DB2 UNLOAD program,
our UNLOADs are getting slower and slower time by time by increasing data in small rate !!!
maybe we are not using a good utility or either it's natural ,
but i 'am sure that IBM utilities should be much better than this
and maybe we are not using the appropriate ones,
for unloading we just create a primary master JCL in our sys1.proclib which accepts some parameters like table name ,region name and so on and uses IKJFT01 utility to copy a data set from one region to another one with simple renaming procedure
but i 've read some article which tells something about
Unload utility which is designed just for this purpose and this utility would bypass the SQL and directly work with table spaces leading to increase the speed.
Would u please give me a JCL sample of these utility or a good source with samples prepared...
also we are running DB2 V7.1 ..
now our jcll is like this ...
//UNLOADT PROC TABLE=,SPC='TRK,(1,300)'
//*
//* UNLOAD DATA FROM A TEST REGION DB/2 TABLE
//UNLOADT EXEC PGM=IKJEFT01,DYNAMNBR=20
//STEPLIB DD DSN=&DB2LOAD,DISP=SHR
//SYSREC00 DD DSN=BKUPUNLD.&SUBSYS..TS&TABLE.(+1),
// DISP=(NEW,CATLG,DELETE),
// UNIT=SYSALLDA,
// SPACE=(&SPC,RLSE),
// DCB=BLKSIZE=0
//SYSPUNCH DD DUMMY
//SYSTSPRT DD SYSOUT=*
//SYSPRINT DD SYSOUT=*
//SYSUDUMP DD DUMMY
//SYSTSIN DD DISP=SHR,DSN=&PREFIX..FNETCOM.DB2DDL
//SYSIN DD DISP=SHR,DSN=&PREFIX..FNETCOM.UNLDCARD(&TABLE)
Joined: 23 Nov 2006 Posts: 19244 Location: Inside the Matrix
Hello,
Yes, the increase in time should be fairly linear. As you add more data, the unload/load will take longer.
Hopefully, when your data was defined/created, there was purge criteria specified also. If your data is forever-to-date, any unload/load process will continually take longer as will many/most of your other processes.
If you post your control statements, we may be able to offer suggestions.
Hi
Remember that always the problem may not be from the utility ,
they are other critical success factor which influence the time
for example in our site ofter a long investigation we found that
the volumes which was dedicated for unload datasets was MIRRORED
like all other volumes in Production system ,
mirroring may not have a sensible increase for small transaction but when you want to take an unload it would effect the total time so much,
mirroring a record of table(for online transactios) is not compatible with mirroing a full data base basckup(all tables) ,
so check the setting of your site through IS to see if the disks which you are going to take UNLOAD are mirrored? if so discard them from mirroring because it is not neccessary, in all site they are taking some other strategies for protecting such data(Daily back up, archiving log data to tape , ext ).
