We are using an OC-12 Sonnet Ring Network <about a 12 gig bandwidth> - Many
of our switch to router configurations are on a 1 gig setup.
To answer your questions, no, the Oracle clients are not in a dedicated
network - they are on the same network with all our other servers. We have
discussed created a dedicated network, but have not gone to far into it at
this time. We have approximately 250 servers from NT to Novell to Solaris -
and at the same time that RMAN is doing its backup at night, the other
servers are backing up as well.
We have been having problems that we are working on resolving with our
network H<we tried many things and have not been able to find the answer -
if anyone has any suggestions for what we can do - that would be great - we
have done many upgrades to our network and also used network monitoring
tools to try to find the bottleneck - but with no luck>
however, the RMAN servers that do about 300gig take approx 5 hours to
complete - before our problems occurred, we were averaging about 3 1/2 hours
- this is with 4 channels. During the day, when RMAN is doing his
archiving, TSM is busy doing its copy pools and reclamations. Therefore,
RMAN is rarely running alone on anything, but it does not run with any
backups during the day which is when the archiving takes place.
The RMAN clients use the ATLP3000 robot that we have. It contains 12 drives
- 2 per each SCSI controller. The TSM is on an E3500 Sun Box that runs
Solaris 2.6.
I hope this information helps and with the server being down today, I sent a
copy of this to your personal box. Let me know if I can be of any more
help.
-----Original Message-----
From: Sean M English/Charlotte/IBM [mailto:[EMAIL PROTECTED]]
Sent: Thursday, October 19, 2000 10:36 PM
To: [EMAIL PROTECTED]
Subject: Re: RMAN question??
Terry,
Thanks for your help. One more question, what kind of backup times are you
seeing and what kind of network are you using? I was just wondering if the
network was a dedicated network for your Oracle clients and are other
backups going on at the same time? Once again, thanks for your help.
Regards,
Sean M. English
IBM Global Services- South Delivery Center, Charlotte, NC
Distributed Storage Management/TSM
External: (704)594-2232 T/L: 8-794-2232
Pager: (800)946--4646 PIN: 1428698
Good news is just life's way of keeping you off balance.
"Barth, Terry (MBS)" <[EMAIL PROTECTED]>@VM.MARIST.EDU> on
10/19/2000 06:19:53 PM
Please respond to "ADSM: Dist Stor Manager" <[EMAIL PROTECTED]>
Sent by: "ADSM: Dist Stor Manager" <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED]
cc:
Subject: Re: RMAN question??
Hi Sean -
We have various RMAN clients using anywhere from 1 to 4 channels - we do
not
have collocation turned on, because of problems such as you mentioned. To
increase our times, we turned off collocation <we have a separate tape pool
for our RMAN clients> and increased channels where necessary. I am not sure
how much your clients back up - but ours back up anywhere from 100gig to
250
gig at night and do archives <which are only about 1 gig> during the day.
In either case, if you do opt to keep collocation turned on, the answer to
your question is yes, it will wait for the media and then start writing the
data as it would if it was not waiting. We have had this situation one time
when we did not realize that the tape pool got low. However, we have found
that RMAN work more efficiently when collocation is turned off and multiple
channels are ran and restore speed was not effected.
Terry
-----Original Message-----
From: Sean M English/Charlotte/IBM [mailto:[EMAIL PROTECTED]]
Sent: Thursday, October 19, 2000 12:03 PM
To: [EMAIL PROTECTED]
Subject: RMAN question??
I have a question about a client that is using RMAN to backup there Oracle
databse to TSM.
TSM Server:
3.7.3.6 on AIX 4.3.2
We recently changed the backup method for the Oracle databases from going
to disk to going directly to tape. We created a unique nodename for each of
the Oracle DB servers. These servers are in there own domain. When we first
tested this, we saw great results (in terms of speed) to backup the Oracle
DB. However, now when the servers run their backups each night, they are in
a constant Receive Wait and the time to back them up as actually increased.
Currently they are allocating one channel for each backup. The reasoning
behind one channel was that anything other than 1 channel (i.e. 2 channels)
the other one would be in a media wait because of colocation. My question
is, has anyone else seen where a backup like this was in a constant receive
wait when using only on channel? I have a thought, but I am not 100% sure
if it would solve my problem because I am not 100% sure on how RMAN groups
its data. My thought is if you run 2 channels, one will be running and the
other waiting. However, will that second channel already have the data
grouped together and ready to transfer when the other one gets finished?
Let me know if anyone else has seen this. If I need to provide more
infomation, let me know. Thanks for the help.
Regards,
Sean M. English
Distributed Storage Management/TSM
IBM Global Services South, Charlotte, NC
External: (704)594-2232 T/L: 8-794-2232
Pager: (800)946--4646 PIN: 1428698
Good news is just life's way of keeping you off balance.
