Wanda,
As always, thank you for your helpful response.
At 01:12 PM 12/21/2005, Prather, Wanda wrote:
I have no experience with ISCSI, so I don't know how well that techology
works with tape drives.
But I have done a campus-distance solution with a remote tape library
connected by fibre.
How remote? Were you still using FC? FCip?
If you can manage a remotely connected library, that is WAY superior
than the other choices.
I can see that, but I'm looking for someone who has done this over
extended distances (i.e., 200 miles) to see how feasible it is.
Drawback of virtual volumes:
Virtual volumes can be messy to manage. When you have IP connection
errors between the two servers, I have not found the messages to be
helpful or easy to resolve. To do a DR recovery, you still need to
rebuild the primary TSM server and have them BOTH up before restoring
clients. And RECLAIMS are especially nasty: to reclaim data, it has to
travel from the remote server, back to the buffer in the primary server,
then back to the remote server.
True, but I don't think this is any more load on the network, just on
the remote server.
Drawback of PTAM:
This is the cheapest solution, for sure.
Do you really think so, over the long run? Staff aren't cheap, and
since we are talking about 200 miles, transportation costs aren't cheap either.
No extra hardware required. A
major drawback is that the tapes you send offsite probably won't be
collocated. When you have a DR situation, you first have to get the
tapes and carry them to your DR site. Then get a TSM server working &
restore the DB. Then it can take a zillion mounts to recover one
client, and you may be very limited by the number of drives available.
Using a remote library:
Assuming you have the connectivity issues worked out, It is VERY easy to
implement. TSM just knows you have 2 libraries. Primary pool in one,
copy pool in the other. TSM doesn't care that they aren't located in
the same building. Tapes collocated in both libraries. In a DR
situation, all you have to do is provide connectivity to a DR TSM
server, reload the data base, and start restoring. We have even ordered
a spare server to sit offsite as a backup Domain controller and stand-by
TSM server. We won't have to restore AD because it will have failed
over to the spare domain controller, and we'll have TSM already
installed, so all we have to do is restore the TSM DB and we're ready to
rock.
In one of the networks we're planning on an all-disk cabinet instead of
a primary tape pool. We're locating the DISK offsite, and the tape
library for the copy pool ONSITE. That way we not only have the data
collocated, we have an unlimited number of virtual drives for restores
in a DR situation!
I've thought of this too. But unlimited drives doesn't mean
unlimited throughput. A lot of the SATA arrays that I've looked at
have throughput limits that could become an issue in a mass-restore
scenario. Still, not having to do all those tape mounts and tape
seeks would save a lot of time.
Again, this is all implemented in an environment where we are within
fibre distances, so I haven't investigated the hardware needed to deal
with your case.
Wanda Prather
"I/O, I/O, It's all about I/O" -(me)
Thanks Wanda.
-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
Paul Zarnowski
Sent: Wednesday, December 21, 2005 12:01 PM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: copy pool architecture question
We are looking to implement a copy storage pool in a distant location
(200 miles away). Network connectivity is good - 1Gb e/n to start
with. DR site is not necessarily the same location as the copy storage
site.
We are considering the following options:
1. use of a remote TSM server and using virtual volumes, or
2. remotely attached tape library using iSCSI to tape drives, or
3. PTAM (pickup truck access method - moving tapes around)
If anyone has experience with options 1 or 2 that they would care to
share, I would appreciate it.
Concerns that I have:
Option 1: using copy tapes for DR would require both primary and
remote servers to be relocated/recreated at DR site, instead of just
the primary server.
Option 2: would like to know if anyone else is doing this and with
what h/w.
Thanks in advance.
..Paul
--
Paul Zarnowski Ph: 607-255-4757
Manager, Storage Systems Fx: 607-255-8521
719 Rhodes Hall, Ithaca, NY 14853-3801 Em: psz1 AT cornell DOT edu
--
Paul Zarnowski Ph: 607-255-4757
Manager, Storage Systems Fx: 607-255-8521
719 Rhodes Hall, Ithaca, NY 14853-3801 Em: psz1 AT cornell DOT edu
| 