Urgent Advice - How to backup large amounts of Data to Disk

[Samuk]

Hi,

At present, we are thinking we may need more of an enterprise product as Backup Exec is not right for our requirements.

The Backup server is a DL580, connected to an EVA6100 with FATA disks to be used as its staging area. Storage is near 15TB. This server has 6 network cards installed.

We have 10 Server we need to backup to disk initially and then duplicate the a SAN attached HP MSL2024 tape library.

2 Servers are attached to the EVA 6100 using Fibre dives and have the largest amount of storage allocated \ used.

Storage Allocation for the larger servers:

Server1:

• 1.95TB
• 1.3TB
• 1.4TB

Server2:

• 1.95TB
• 1.5TB

Backup Exec Configuration:

Selection Lists

3 Selection Lists created as follows:

• List 1: Server1
• List 2: Server 2
• List 3: Smaller Servers with local storage

Policies

A single policy has been created to backup the 3 selection lists to disk and duplicate to Tape.

Issues

We are seeing poor performance when backing up to disk. I did a backup of justthe 1.4TB server disk results: I did a backup to tape and a seperate backup to tape to get the speed \ time differences.

• Backing up to disk: 14 hours
• When backing this up directly to tape: 7 hours

This will ofcourse exceed the backup Windows if i were to backup everything to the staging area.

Does anyone have any thoughts..

Appriciate any help.

Sam

[CraigV]

Hi Sam,

So what I see is the following:

1. Backup server connected to EVA 6100, backing up a multitude of other servers. Runs through an FC switch right?

2. 2 larger servers connected to the EVA with direct-attach to the host ports.

3. The rest backing up other data.

Going on that, even if you have disk presented to the media server, your backups are going to run across the LAN. the reason for this is that you're using the RAWS agent on those remote servers. It doesn't matter what storage you were direct-connected too, RAWS always runs across the LAN. So this explains (in part) the slow speeds you're getting. With FATA drives you also can't expect the same sort of performance as you would get from FC drives (10K/15K), but they would do the job as they're low-end disk specifically for this reason.

There is a way around this, but there is added cost involved. The answer might lie in licensing Backup Exec's SAN SSO option. This allows SAN-attached servers to back up to SAN-attached devices (tapes/disk). The downside here is that you need to license the SAN SSO license, and then a full installation (with agents) of Backup Exec 2010 to take advantage of this.

1 other option might be to split those jobs up into 3, and not have just 1 job doing the work. This allows multiple streams of data to the B2D, theoretically speeding up your backup times. Therefore 1 job per selection list, and also increasing the number of simultaneous jobs to the B2D.

If you want to read up on SAN SSO, best bet is to get hold of the BE 2010 Admin Guide and read the relevant chapter.

I have used it in the past, and it cuts backup times down substantially. This was to an LTO2 library though...

Craig

[Samuk]

Hi Craig,

well we have a full version of Backup Exec 2010, and Agents.

We are in the process of buying the Advance Disk Option, can this be run in parallel with the SSO?

I did look at the SSO a few weeks back but was mis-informed by the suplier as to what it is for! I guess i should have read up on it!

I am looking into the costings for this - and am going through the user guide to see how this works. Is it complicated to use \ setup?

Can i use this as a trail even if my Backup Exec is not in trail mode?

Quesiton i guess is will the SSO help or is this something i will find out after trying?

The MS2024 library is a LTO4 - but i need to get the disk based backups to speed up first.

About the jobs:

Do you mean split up the selection lists for the large servers down further? as currentyl 3 jobs are running to the same B2D folder

[CraigV · Accepted solution]

OK...let me make some clarifications here:

1. The SAN SSO is for tape devices primarily. It means you can share your tape library with other SAN-attached servers.

2. It is relatively easy to set up, but you need a full installation on each of the other servers.

3. You can share your library amongst your servers, but still carve out disk space from the EVA 6100, and assign it to each of those servers. From there, you can then kick off a duplicate to tape job. This way a job runs on each server, freeing up the media server to back up the other servers. And if need be, you can configure this server as a CASO server to manage the other 3 SAN-attached server jobs.

Thanks!