Networker

Re: [Networker] Networker D2D2T backups and D2T cloning speed

2008-08-07 07:02:58
Subject: Re: [Networker] Networker D2D2T backups and D2T cloning speed
From: Francis Swasey <Frank.Swasey AT UVM DOT EDU>
To: NETWORKER AT LISTSERV.TEMPLE DOT EDU
Date: Thu, 7 Aug 2008 06:55:47 -0400
On 8/7/08 1:11 AM, Mitko Stoyanov wrote:
* When Networker clones from adv_file device to tape, is the adv_file
saveset copied as it is without looking inside what is saved, or is
"restored" internally and then backed up as a saveset to the clone tape?
The answer to this one has to be that the saveset is copied without looking inside from the adv_file to the tape. I don't know exactly how NetWorker performs this, but I know that I don't have the 30TB available to networker to do the restore and then back it up. We also all know that NetWorker doesn't do cross-platform restores, so if NetWorker needed to restore and then backup to do a clone, it wouldn't be able to clone any clients that were not the same platform as the server doing the cloning.

* Is there any workaround for the single-threaded saveset cloning? Any
plans to multiplex the adv_file savesets when cloning to tape? That
looks to me as a major showstopper when using faster tape technologies
to clone D2D backups, as eventually these backups have to go offsite,
and the vast majority of clients use tapes for that..
No, there is no method to get NetWorker to multiplex a clone tape (that I have found yet). What I have found in a practice of doing just such a setup with D2D2T using both Nexsan SATABeast and DataDomain DDR adv_file devices and a Qualstar XLS with LTO4 drives that the cloning from disk to LTO4 completes faster than it ever will from tape to tape. Therefore, even though there are times when the LTO4 drive is not being driven as fast as I would like (shoe-shining) the overall average speed of the clone is higher than if NetWorker was searching for the saveset fragments on a highly multiplexed backup tape.

There was a brief time when I was having systems back up to LTO4 and then clone from LTO4 to LTO4. For the systems that are now backing up to disk, I have seen the clone time reduced from three days to one day. That tells me that the disk investment was well worth it. I can actually almost depend on having Thursday and Friday when the system is not busy and I can take it down to do tests for EMC on my open issues. (The reason I can't depend on it is I have one large customer that refuses to foot the bill to do disk -- their backup tapes are so highly multiplexed that their clones take forever).

I also really like that the clone process does not multiplex the data onto the tapes, as it will make the restores from the clones require fewer tapes and also go faster.

Imperically, it appears that given a list of saveset id's to clone, that NetWorker processes them one at a time. Thus, for a list of small savesets, you'll find that NetWorker spends more time setting up to do the copy than it does doing the copy -- and you'll see that in NMC as a low write speed to the LTO4 media. At least, that's what I've convinced myself is happening...

--
Frank Swasey                    | http://www.uvm.edu/~fcs
Sr Systems Administrator        | Always remember: You are UNIQUE,
University of Vermont           |    just like everyone else.
 "I am not young enough to know everything." - Oscar Wilde (1854-1900)

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