DOH!!! While I was typing this I remembered that I was thinking of in terms of NDMP backup performance was QTree vs Volume, not Volume vs named snapshot. Your named snapshot is essentially a volume level backup so it should not be signifcantly different in performance. QTree backups are much slower because it has to determine which files belong to the QTree then which blocks those files use.

On Wed, May 9, 2012 at 2:45 PM, John Stoffel <john.stoffel@taec.toshiba.com> wrote:

Bill> When you start a volume level NDMP backup, Data ONTAP takes a
Bill> snapshot, does its backup from that snapshot, then deletes that
Bill> snapshot. If the data is basically static at the time you wish
Bill> to perform your backup then you could take your named snapshot
Bill> for faster recovery purposes and immediately thereafter start
Bill> the volume level NDMP backup - assuming you have available
Bill> destination devices.

That's the kicker, I probably *won't* have enough available tape
drives to run all the volumes in parallel, since I need to make sure
my backups (snapshots) are taken close in time.

Bill> Otherwise, what you have described will likely give the best
Bill> performance.

Can you give an idea of show slow NDMP backup of a named snapshot is
versus a regular NDMP backup using it's own snapshot? 25% slowdown?
50%? I'm running on a FAS3140 which shouldn't too loaded at all.

I'm also working within the limitations of CommVault 9, so I may not
be able to specify a named snapshot to backup from there anyway.

Thanks for all your feedback.
John

Bill> On Wed, May 9, 2012 at 2:33 PM, John Stoffel
Bill> <john.stoffel@taec.toshiba.com>wrote:

>>
Bill> Even if it is possible, I would recommend against it.
Bill> Performance of an NDMP backup from a named snapshot is
Bill> significantly slower than a volume level NDMP backup.
>>
>> Blech, not good to hear.
>>
Bill> Are you performing your backup directly from your source volume
Bill> or a snapvault secondary volume?
>>
>> Ideally, directly from the source volumes, plural. I'm going to need
>> to backup the volumes from a snapshot time that's as close as possible
>> to the same time, then I can take my time doing the actual NDMP to
>> tape of the data.
>>
>> So it's looking like my best plan would be to just do:
>>
>> foreach v (data_a data_b data_c)
>> snap create $v flexclone_backup
>> vol clone create ${v}_clone_bkp -s none -b ${v} flexclone_backup
>> done
>>
>> then backup my new data_a_clone_bkup... volumes via NDMP.
>>
>> John
>>
>>
Bill> On Wed, May 9, 2012 at 1:53 PM, John Stoffel
Bill> <john.stoffel@taec.toshiba.com>wrote:
>>
>> >>
>> >> Guys,
>> >>
>> >> Anyone know if it's possible do an NDMP backup (using CommVault) of a
>> >> *named* snapshot? Basically I need to snapshot two or more volumes at
>> >> the same time (I can do this in a pre-script) and then let the
>> >> snapshots be backed up.
>> >>
>> >> I guess I could be silly and just create a flex clone volumes and then
>> >> back those volumes up instead, which would get rid of the issue, but
>> >> would complicate things a bit more.
>> >>
>> >> Thanks,
>> >> John
>> >> John Stoffel - Senior Staff Systems Administrator - System LSI Group
>> >> Toshiba America Electronic Components, Inc. -
>> http://www.toshiba.com/taec
>> >> john.stoffel@taec.toshiba.com - 508-486-1087
>> >> _______________________________________________
>> >> Toasters mailing list
>> >> Toasters@teaparty.net
>> >> http://www.teaparty.net/mailman/listinfo/toasters
>> >>
>>