Hi, I have 4 TB db2 db. all tablespaces complete very fast except one tablespace PSAPBTABD (2 terra byte )

** This tablespace takes 36 hours to completes
** The start backup throughput is 100MB / second which is good
** after 3 hours the backup throghput falls down to 10MB / sec which is bad

We have tried the following:

** Increase utility memory
** rebalance the PSAPBTABD from one stripe set to 10 stripe sets
** Increase the DB2_PARALLEL_IO=*:5
** Upgrade the DB2 9.1 to latest fix pak 10

Nothing works, this tablespace is very slow, all other small tablespaces take good backup throughput of 100mb/sec. We have called in the TSM tivoli vendor. He confirmed there is no issue with TSM layer

Anyone has any thoughts?

Thanks in advance...

Specify multiple paths for the backup, preferably on different disks or different disk arrays (each with its own mount point). For example:

backup db sample to /db2/backup1/sample, /db2/backup2/sample, /db2/backup3/sample, /db2/backup4/sample compress

Hi, I am using TSM with 4 sessions, also tried with 8 sessions and parallelism 16.

The initial 20% of the backup goes very fast and after that It slows down and takes 36 hours

What do you mean by stripe set? You have 10 Containers for the TBS? Or do you have a single LUN build from 10 Stripe Sets from your SAN?

100 MB/s is compressed or not? If it's uncompressed, i'ld guess there's a bottleneck somewhere (network???). But that's more general and not related with the backup of your tablespace.