I have a 340 gig db with a 290 gig lob file. The backup takes 11 hours. I was wondering what sort of hardware and software alternatives I have for doing faster backups. I found the split mirroring backup routine from the documentation, but not sure how that would be setup. Also, I wish someone from the business area would say what sort of (high) availability they need, but alas I am sure they will say it cannot be down for 1 minute.

What OS is the database server? What are you backing up to?

Split mirror requires special hardware/software on the server where the database is stored. It is basically making a super-fast (hardware) copy of the hard disk holding the database. The disk hardware has to be able to do this.

Andy

ahhh, Windows 2003, and probably eventually Windows 2008 R2.

What are you currently backing up to?

Andy

It's going to a SAN drive now.

Is the database on the same SAN? How big is the backup image?

Andy

The db is actually on local disk still (340 gig) (total disk is 700 gig), the backup goes to the SAN and is compressed down to 245 gig. It grows about 1 to 2 gig a week.

Hi,

for Split Mirror or Snapshot Backups the database has to be placed on SAN, too. These Functions have to be supported from the SAN-Storage aswell.
What is your bottleneck? Are you CPU-Bound or I/O-Bound?

How long does the backup take if you do not do the compression?

Andy

I never tried to do it without compression.

What version of DB2 and fixpack are you using? What is your server like (CPUs and memory)?

Andy

v8 FP 12. 3 gig RAM, Intel Xeon 3.2 ghz (total of 4), Windows 2003 Standard Edition. I have no idea on number of drives and how it's configured, it is some form of SCSI though.

First V8 went out of service over a year ago. You can no longer get any support on it. You really should think about upgrading to one of the V9 version.

Second, compressing backup images is server resource intensive. You have limited memory, so you may have a bottleneck there. See how long a backup without compression takes.

You can do online backups that allow users to continue processing while the backup is in progress.

Andy

Yes, v9 is in the planning stages.

Re: online. Yes, I found a table that had not logged on it and need to unload, drop, recreate and reload. Someday we'll get it all done.

Back on topic: We do not care about IBM service strategies, we care about DB2. So lets try to make this backup faster:
Is it an online or offline backup? If online: how is the system load during the backup period? Any batches running the same time? Updates going on? Look at db2diag.log for any logfile archiving going on inbetween your start and end of the backup.
Is the UTIL_HEAP_SZ configured to max? If not do so, you might even borrow same memory from bufferpools during the backup.
You backup to SAN disk. Only 1? So the traffic is using only 1 NIC or HBA. Try to get/use more connections between your server and SAN and backup to more disks/LUNs simultaneously and make sure the traffic is spread over all the connections.

If that doesn not help: monitor your server uring the backup. There must be a bottleneck somewere. Find it and report back to us.