Welcome! » Log In » Create A New Profile

[EMC-DataProtection-L] Cloning speed?

Posted by Anonymous 
[EMC-DataProtection-L] Cloning speed?
November 01, 2016 04:07PM
Perhaps an obvious question, but when cloning save sets, what causes the
read/write speeds to change? Is there a primary culprit?

Let's assume that you're using device compression, not software, and
you're writing directly to tape, and the drives are
directly attached to the server.

I'm unclear on what's really happening here or what the primary
influences are? Is it similar to backups wherein the
compressibility of the data will determine how efficient the tape write
buffer is filled, and thus affect speeds? Moreover,
if you're cloning all the save sets that were multiplexed together, when
they were first backed up, then it preserves that
and doesn't need to de-multiplex it (right term?), correct? If so, then
why would the write speed ever differ? Let's say you're
cloning all the save sets on tape A and all the save sets on tape B, and
you create a list (file1) of all the ssids, and you run
`nsrclone -s server -b 'clone pool' -S -f file1`. It's going to optimize
the order that it does them in, but since it has to read all
the save sets on both tapes, as opposed to a subset, then what causes
the read and write speeds to vary if the network
isn't involved? I don't see write speeds remain consistent for very
long, even when cloning large save sets.

Maybe the better question is if cloning is no different than backups, in
terms of the vagaries of speed, then why or why not?

George

--
George Sinclair
Voice: (301) 713-4921
- The preceding message is personal and does not reflect any official or unofficial position of the United States Department of Commerce -
- Any opinions expressed in this message are NOT those of the US Govt. -

--
This list is hosted as a public service at Temple University by Stan Horwitz
If you wish to sign off this list or adjust your subscription settings, please do so via http://listserv.temple.edu/archives/emc-dataprotection-l.html
If you have any questions regarding management of this list, please send email to owner-emc-dataprotection-l < at > listserv.temple.edu
[EMC-DataProtection-L] Cloning speed?
November 02, 2016 11:40AM
In regard to: [EMC-DataProtection-L] Cloning speed?, George Sinclair - NOAA...:

[quote]Is it similar to backups wherein the
compressibility of the data will determine how efficient the tape write
buffer is filled, and thus affect speeds?
[/quote]
I believe that's part of it. Even on a NetWorker server or storage node,
other things are happening on the system that might cause I/O variances.

[quote]Moreover,
if you're cloning all the save sets that were multiplexed together, when they
were first backed up, then it preserves that
and doesn't need to de-multiplex it (right term?), correct?
[/quote]
It is possible to get NetWorker to de-multiplex savesets on a clone
volume, but assuming you're cloning all of them at once, then it should
be preserving the multiplexing. You may want to review the mailing list
archives on clone by volume vs. clone by ssid too.

There are things that mminfo can report that will give you a relative
idea how savesets are positioned on tape -- look at some of the attributes
you can report on near the bottom of the list. You might even be able to
build a report or graph comparing two tapes, to see if the saveset
layout is roughly the same on the clone volume.

Tim
--
Tim Mooney Tim.Mooney < at > ndsu.edu
Enterprise Computing & Infrastructure 701-231-1076 (Voice)
Room 242-J6, Quentin Burdick Building 701-231-8541 (Fax)
North Dakota State University, Fargo, ND 58105-5164

--
This list is hosted as a public service at Temple University by Stan Horwitz
If you wish to sign off this list or adjust your subscription settings, please do so via http://listserv.temple.edu/archives/emc-dataprotection-l.html
If you have any questions regarding management of this list, please send email to owner-emc-dataprotection-l < at > listserv.temple.edu
[EMC-DataProtection-L] Cloning speed?
November 02, 2016 12:34PM
On 2016-11-02 14:35, Tim Mooney wrote:
[quote]In regard to: [EMC-DataProtection-L] Cloning speed?, George Sinclair -
NOAA...:

[quote]Is it similar to backups wherein the
compressibility of the data will determine how efficient the tape write
buffer is filled, and thus affect speeds?
[/quote]
I believe that's part of it. Even on a NetWorker server or storage node,
other things are happening on the system that might cause I/O variances.

[/quote]Right.
[quote][quote]Moreover,
if you're cloning all the save sets that were multiplexed together,
when they were first backed up, then it preserves that
and doesn't need to de-multiplex it (right term?), correct?
[/quote]
It is possible to get NetWorker to de-multiplex savesets on a clone
volume, but assuming you're cloning all of them at once, then it should
be preserving the multiplexing. You may want to review the mailing list
archives on clone by volume vs. clone by ssid too.

[/quote]I guess I was more curious about why the speed might vary even if you
are cloning all the save sets at once, and not a subset.
I would expect it to go slower if it has to de-multiplex, but if it
doesn't then would other things happening on the system
solely account for the variances or does cloning itself (even if it's
not de-multiplexing) always vary due to its nature, and if
so why.

[quote]There are things that mminfo can report that will give you a relative
idea how savesets are positioned on tape -- look at some of the
attributes
you can report on near the bottom of the list. You might even be able to
build a report or graph comparing two tapes, to see if the saveset
layout is roughly the same on the clone volume.
[/quote]Right. I'll take a look. Thanks. :)
[quote]

Tim
[/quote]

--
George Sinclair
Voice: (301) 713-4921
- The preceding message is personal and does not reflect any official or unofficial position of the United States Department of Commerce -
- Any opinions expressed in this message are NOT those of the US Govt. -

--
This list is hosted as a public service at Temple University by Stan Horwitz
If you wish to sign off this list or adjust your subscription settings, please do so via http://listserv.temple.edu/archives/emc-dataprotection-l.html
If you have any questions regarding management of this list, please send email to owner-emc-dataprotection-l < at > listserv.temple.edu
Sorry, only registered users may post in this forum.

Click here to login