Networker

Re: [Networker] Oracle RMAN backup questions

2006-11-30 17:22:08
Subject: Re: [Networker] Oracle RMAN backup questions
From: George Sinclair <George.Sinclair AT NOAA DOT GOV>
To: NETWORKER AT LISTSERV.TEMPLE DOT EDU
Date: Thu, 30 Nov 2006 17:14:13 -0500
My bust. What I think I meant was that increasing the channels will increase the number of save sets that are *simultaneously* written to tape, but not the total number. The total number will not change, only the filesperset will change this. BUT, increasing the number of simultaneous
save sets being written to tape would increase recover time, right?

Guess I'm trying to understand how filesperset would affect recover time.

George

Doug Brown wrote:

George

Your statement

Clearly, the more channels you set, the more save sets you end up with and thus the more save sets will be interleaved to the tape and the longer recovers will take, so there's a trade off between increased write speeds versus recovers. BUT, the more save sets you break the Oracle backups into the smaller those save sets, and that could be a good thing, but the more filesperset, the larger those savesets, too. Hmm ... OK, all fine and well, but:

is incorrect.

The less files allocated per channel will result in more savesets. By default Oracle will try to divide the number of files into the number of channels allocated as from the oracle whitepaper tuning recover manager ( RMAN )

Backup Multiplexing
One channel can simultaneously read more than one file. The level of multiplexing is the number of files read simultaneously on a single channel and then written to the
same backup piece. The degree of multiplexing depends on the FILESPERSET
parameter of the BACKUP command as well as the MAXOPENFILES parameter of
the CONFIGURE CHANNEL or ALLOCATE CHANNEL commands.
The number of files in each backup set is the lesser of FILESPERSET and the number of files read by each channel. The level of multiplexing is the lesser of
MAXOPENFILES and the number of files in each backup set. For example,
assume that you back up two datafiles with one channel. You set FILESPERSET
to 3 and set MAXOPENFILES to 8. In this case, the number of files in each
backup set is 2 (the lesser of FILESPERSET and the files read by each channel),
and so the level of multiplexing is 2 (the lesser of MAXOPENFILES and the
number of files in each backup set).
Assume a different case in which each channel reads fifteen datafiles,
FILESPERSET=10, and MAXOPENFILES=8. You can calculate the level of
multiplexing as follows:
min(min(15, 10), 8) = 8
The default values are 16 for MAXOPENFILES and 64 for FILESPERSET.





George Sinclair <George.Sinclair AT NOAA DOT GOV> Sent by: EMC NetWorker discussion <NETWORKER AT LISTSERV.TEMPLE DOT EDU>
11/30/2006 01:09 PM
Please respond to
EMC NetWorker discussion <NETWORKER AT LISTSERV.TEMPLE DOT EDU>; Please respond to
George Sinclair <George.Sinclair AT NOAA DOT GOV>


To
NETWORKER AT LISTSERV.TEMPLE DOT EDU
cc

Subject
[Networker] Oracle RMAN backup questions






Hi,

Several questions here on RMAN Oracle backups via NMO. We're running Oracle 10g, with NW 7.2.2.
Our drive target sessions are set to 5. These are SDLT 600 drives.

Some of these questions may be moot since there's so many variables that will determine what settings work best for your shop but really just wanted to get an idea of what others are doing and what, if any, bad experiences,
pit falls or suggestions you could offer.

In testing Oracle backups and recovers, we've been playing with a rather small test database. We've tried multiple channels, and multiple filesperset, and we don't really notice any major performance difference, but obviously, the more channels, the more concurrent streams we can send to the device, so it's more efficient and doesn't have to continually stop and then back up another like it would if we had channels set to 1 where it would go one at a time.
Also, increasing filesperset helps in this area , too.

Clearly, the more channels you set, the more save sets you end up with and thus the more save sets will be interleaved to the tape and the longer recovers will take, so there's a trade off between increased write speeds versus recovers. BUT, the more save sets you break the Oracle backups into the smaller those save sets, and that could be a good thing, but the more filesperset, the larger those savesets, too. Hmm ... OK, all fine and well, but:

1. Does more filesperset also affect recover time in the same way that more save sets (channels) to the same tape does?

2. During normal (non-Oracle) backups, how many files does NetWorker send in a single saveset?

3. Let's say you had a database of 100 files, maybe 70 GB total. If you set filesperset 1, and had 1 channel then you'd end up with 100 save sets on a full which sounds kinda dumb. On the other hand setting filesperset 100 with 1 channel sounds ridiculous as well. So, how about 2 channels with filesperset 5. That would take 20 save sets on a full, maybe only 5-6 save sets on an incremental. Does that sound reasonable or would it be better to crank it up to 5 channels? Our devices are set to handle 5, so it's not gonna need another drive.

We have the Oracle client's parallelism set to default 4 (NetWorker), so I imagine we can't increase the channels beyond 4 unless we change that value. This machine has 8 cpus, plenty of memory, so maybe we can increase that. Since our test database is so small, though, we really don't have a good feel for what will and won't work so well, but I was thinking that if we're seeing decent write speeds, and decent recover times at say 2 channels with
filesperset 5 then maybe just leave it at that?

4.  Should the channels be set to 1 when recovering?

5. Should we use regular differential backups (like level 1 every night until the next full) or a cumulative backup?

Thanks for any horror stories or suggestions.

george



--
George Sinclair - NOAA/NESDIS/National Oceanographic Data Center
SSMC3 4th Floor Rm 4145       | Voice: (301) 713-3284 x210
1315 East West Highway        | Fax:   (301) 713-3301
Silver Spring, MD 20910-3282  | Web Site:  http://www.nodc.noaa.gov/
- Any opinions expressed in this message are NOT those of the US Govt. -
To sign off this list, send email to listserv AT listserv.temple DOT edu and type 
"signoff networker" in the body of the email. Please write to networker-request 
AT listserv.temple DOT edu if you have any problems with this list. You can access the 
archives at http://listserv.temple.edu/archives/networker.html or
via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

<Prev in Thread] Current Thread [Next in Thread>