- Posts: 6
- Liked: never
- Joined: Sep 13, 2010 10:34 am
- Full Name: Harry Leak
I am extremely new to Veeam, and would appreciate some assistance with getting my head around configuring the backup jobs in the best way to suit my infrastructure.
We currently have....
2 x ESX 3.5 Servers in VIC Cluster connected to..
2TB fibre channel SAN
1 x Standalone ESX 3.5 host
LTO3 tape library
A physical VCB Server controls the backups and has a LUN allocated to it from the SAN, totalling 270GB in size (was previously used with Symantec Backup Exec & VCB mount/dismount) Symantec Backup Exec also resides on the same VCB server of which I have Veeam Backup & Replication installed.
We have a mixture of servers, some of these need to be backed up daily and some can be done weekly. We also run Exchange 2007 and 3 x SQL 2005 servers.
I am aware that "changed block tracking" although enabled will not work until we upgrade the hosts to version 4.0 or above, of which we have scheduled some downtime in to do this early next month as we are a production based business so it is difficult to schedule it in to be completed.
To create a full backup of everything would take up nearly 2TB's so I have the use of a 2TB NAS storage device which is accessible from the VCB server which I have been using for test backups currently.
My initial thoughts as disk space could be an issue with storing multiple rollbacks purely on disk was to create daily and weekly backups of the VM's in question including all disks to the NAS device, and include one full backup per week on a Saturday when production was not going on with less rollback points. I was then going to either create a job manually in Symantec Backup Exec to backup the reversed incremental VBK file to tape, or by using the post command to initate the job immediately after the Veeam job has finished. This would then give me archived backups as I would be unable to store that many rollbacks on disk.
The questions I have are as follows
1. - Until I upgrade to ESX 4.1, should VStorage API be used for the backup job(s) as performance could still be an issue.
2. - Is it best create one job per VM, or do you include them all or group similar servers? Does this make scheduling a bit of a nightmare if they are all seperated?
3. - Is there a better way of doing this with the hardware and available disk space and resources I have available?
4. - Should I create seperate jobs for backing up the Server C: drives (less regularly) and the remaining data drives more frequently?
5. - Does Veeam do an application level backup of Exchange and SQL or should I still use Symantec Backup exec agents to perform these?
I'm just really after some "best practice" advice on configuring Veeam to work in my environment - Any assistance from you would be greatly appreciated.
Thanks for looking.
- Product Manager
- Posts: 24351
- Liked: 1894 times
- Joined: Mar 30, 2009 9:13 am
- Full Name: Vitaliy Safarov
1. Provided you have direct FC connection to your SAN from the VCB proxy server, I would choose vStorage API - SAN mode to run the backup jobs. It should give you best performance possible.
As regards to CBT... that's correct, it is only available for vSphere hosts, though vStorage API - SAN mode will work with 3.5 hosts as well.
2. In order to get best deduplication ratios, you should include all similar VMs (made from one template, or having the same OS installed) in one job, as deduplication doesn't work across other jobs.
3. If best setup practices are concerned, I would recommend to have daily jobs with retention policy set to 5 or 7 in order to cover all business days for your mission-critical servers such as Exchange and SQL, then each saturday/sunday I would execute a BE job to offload backups (VBK) to tape. For all other VMs I would use weekly configured backup jobs with post backup job script option enabled to offload backups to tapes immidiately after the job finishes.
While using synthetic way of backup, you don't need to create Full backups, unless you have some corporate policies to follow, that would allow you to save some free space on your target storage.
By the way, we have a whitepaper describing Veeam Backup and Replicaton/Backup Exec integration, might be helpful. This document is available through our support team.
4. Actually it depends, if you're backing up Exchange or SQL server I would definitely choose entire VM (without disk exclusions). But taking into consideration file servers, where the most valuable disks are non-system disks, I would probably choose to exclude system disk from backup job.
5. Veeam performs image-level backups (whole VM), thus meaning single pass is theoretically all what is needed to backup everything inside the VM. Furthermore, with U-AIR functionality coming with v5 soon, you will able to restore single items from those applications from the given entire VM image.
Hope it helps!