oracle tdp seems to be doing nothing

[wesd]

Hi,

I've configured TDP for Oracle 6.3.0.0 on an aix 6.1 client. The tsm server is also 6.3.0.0 on aix 6.1.

When I use rman to test a backup, I can see a session starting on the tsm server but very little seems to be hapening. A query session shows -

tsm: BCHTSMLIVE1>q sess

Sess Comm. Sess Wait Bytes Bytes Sess Platform Client Name
Number Method State Time Sent Recvd Type
------- ------ ------ ------ ------- ------- ------- -------- ----------------------------
146,110 Tcp/Ip RecvW 10 S 3.6 K 24.0 M Node TDP Oracle AIX ECRDEV_ORA

The figure for bytes recvd is increasing incredibly slowly - the 24M figure has taken 30 minutes to reach - while the rman ouput seems to be stuck at -

channel t1: starting piece 1 at 29-AUG-12

The entire database is only 17GB so I would expect it to complete very quickly.

The tsm server is configured to send the backup to a disk storage pool which uses ibm ds8100 disks (ie. very fast san-based disk).

The management class is being picked up ok - if I change the copy pool destination from disk to tape and start another rman backup the tsm server immediately mounts a tape.

Has anyone seen this behaviour before?

 

[wesd]

I've just tried the backup with a single database file - 1835008 bytes backed up in 00:01:43 with a throughput rate of 17.40Kb/Sec.

The configuration seems to be working, at least, but very slowly. I'm also a bit worried about the 1835008 bytes - the datafile I picked is 104865792 when listed in unix using ls -l. I don't know how oracle allocates/uses space with its database files but does this sound normal?

 

[BBB]

File size could be related to compression.

What kind of network interface is it? Can you ftp (put and get) a ~10mb file between the tsm server and client to test network?

 

[Jeroen]

The TDP for oracle will only backup the actual data in the database, not the database file itself that exists on Unix.

 

[wesd]

I don't think network speed is an issue.

I ftped the datafile I saved in tsm - approx 100MB - with the results below.

ftp> put wdtar
200 PORT command successful.
150 Opening data connection for wdtar.
226 Transfer complete.
104867840 bytes sent in 1.056 seconds (9.702e+04 Kbytes/s)

ftp> get wdtar
200 PORT command successful.
150 Opening data connection for wdtar (104867840 bytes).
226 Transfer complete.
104867840 bytes received in 1.112 seconds (9.206e+04 Kbytes/s)

A different file of 1GB in size -

ftp> put wdtar
200 PORT command successful.
150 Opening data connection for wdtar.
226 Transfer complete.
1073756160 bytes sent in 11.27 seconds (9.305e+04 Kbytes/s)
local: wdtar remote: wdtar


ftp> get wdtar
200 PORT command successful.
150 Opening data connection for wdtar (1073756160 bytes).
226 Transfer complete.
1073756160 bytes received in 12.04 seconds (8.707e+04 Kbytes/s)
local: wdtar remote: wdtar

Perhaps oracle rman has a lot of processing to do although it's a development server so the idle time is > 90% with very little disk activity.