So has anyone actually tested the restore speed of large volumes using DD
and Boost?  What have been your results?

We have sales people beating down our doors from IBM - TSM trying to pitch
that the incremental forever is the solution for everything.  I just dont
see it.  We currently have Networker 7.6.3 and in the process of migrating
clients over to 8sp1 and so far things have been working good.  We do use
traditional tape(LTO4) in our environment so we have the normal issues that
come with backing up to tape but outside of that things work pretty good.
I would love to see what a DD poc would do for us and how it would change
our environment in general.  Also, I mentioned that we could have client
direct backups which would also speed things up but then was told that we
wouldn't want to push more onto our clients as far as overhead, but I would
have to assume that the overhead of a client direct client has to be less
than that of a traditional client backup having to send over all of the
data.

Let me know your thoughts on this and any other info that I am leaving out.
Also, I would be interested if any of you have migrated from IBM to
Networker.

Thanks,

Tony


From:   Tim Mooney <Tim.Mooney AT NDSU DOT EDU>
To:     NETWORKER AT LISTSERV.TEMPLE DOT EDU
Date:   04/08/2013 11:51 AM
Subject:        Re: [Networker] Data Domain in mixed BOOST/VTL mode?
Sent by:        EMC NetWorker discussion <NETWORKER AT LISTSERV.TEMPLE DOT EDU>



In regard to: [Networker] Data Domain in mixed BOOST/VTL mode?, bingo
said...:

> Why do you really want to use VTL at all?

I didn't say I wanted to -- in fact, it would be my preference that we
didn't.  However, our SAN folks are advocating the VTL attachment for
the clients that have SAN storage.

> Using a DD you are talking to very fast network-attached AFTDs where NW
> 8 clients can even use new technologies like Flexible Blocksize & Direct
> File Access. The data rates which you achieve here are amazing,
> especially when you stream data from various clients to multiple AFTDs
> and - at the same time - simulteanously clone data from these devices to
> tape (if needed at all).

We understand all that, but I think you're focused on the backup
aspect.  If we have to do a full restore of a 1.8 TB volume, all the
data still has to be transferred from the DD to the client.  Client direct
and de-dupe isn't going to help us there.

So, in theory (hah!) it should be much faster to restore a large volume
over our 4 GiB SAN than it would be to pull it all back over a 1 GiB
TCP/IP link.

> My opinion is to avoid VTLs whenever possible. Yes, you have fast SAN
> access but IMHO this is the the only benefit. You really force a disk to
> behave like a tape and in general, you add an additional, unnecessary
> 'layer' which limits the flexibility B2D would offer. And of course it
> is a an additional component which potentially might cause trouble. Just
> avoid it!

I'm not disagreeing with you Carsten, but unless there are some
significant gotchas with using a DD in mixed mode, I don't think any of
that reasoning is going to hold out against the potential for greater
restore speeds.

Tim
--
Tim Mooney                                             Tim.Mooney AT ndsu DOT 
edu
Enterprise Computing & Infrastructure                  701-231-1076 (Voice)
Room 242-J6, IACC Building                             701-231-8541 (Fax)
North Dakota State University, Fargo, ND 58105-5164