Well if you have got to this page you probably already have backups running to disk (and duplicate to tape) via the VCenter using AVVI and your looking to turbo charge your backups.
Well here are the simple truths.
- You muck this part up and the windows host initialises the LUN that your datastore is on… there goes all your VMs. You have been warned
- Getting Backup Exec to use the SAN is straight forward. Obviously your VMhosts and BE Media Center must be connected to the SAN via HBAs. The VCenter does not have to be, it is just acting as a “go between”
- If your backing up directly to tape, the only gain you are going to get is likely to be marginal as the tape drive is likely to be your bottleneck. But putting the backup traffic over the SAN switching fabric will free up bandwidth on your main network.
- If you are backing up to disk, then prepare for some major reductions in backup times. With the SAN infrastructure I have been using we are seeing more than a 5 times speed improvement (from 3.5GB/min to 20GB/min)
- If you’re doing B2D and duplicate to tape then just add items 2 and 3 together.
- To use a couple of sayings:
- There is no substitution for cubic inches – If you have powerful kit, you will get good results.
- You can’t make a silk purse out of a sow’s ear. – if you have poor performing kit, don’t expect this to make it faster.
- Basically – try anywhere other than windows to present the LUN as read only. This recommendation is because it can be too easy for some (less than bright) sys admin to wonder why the disks are in the server and not in use and make them in use and assign a drive letter, thus wiping the disks that contain the VMs!!!! By making the
If your SAN supports presenting shared LUNS as read only to one host (media server) and read/write to the other(ESX VM host) then do the read only limit there, rather than in windows
If your HBA in the media server can limit the LUN to be read only, then do it there, rather than in windows
FYI – the kit I have used to get this speed gain was:
SAN IBM V7000
SAN switching fabric IBM 8Gb/s
HOST IBM x3650 M3 6C X5650 2.66GHz 12GB HS SAS but with ESX4.1 booting off USB Stick
Media server IBM x3550 M3, Xeon 4C E5620 80W 2.40GHz/1066MHz/12MB
Tape Library IBM TS3200 Tape Library. Single tape head ULT3580-HH5 connected to SAN
Backup Exec acts as a VCB Proxy, so follow the general instructions on pages 33 – 35 for setting up VCB Proxy
http://www.vmware.com/pdf/vsphere4/r40/vsp_vcb_15_u1_admin_guide.pdf
In my case this is what I did:
- Shut down the Media Server (Power off the server)
- Unmap the LUNs for the host in the Integrated Management Module of the SAN
- Boot the media server and log in as administrator.
- Open a command‐line interface.
- Run the Diskpart utility by typing:
diskpart
The Diskpart utility starts up and prints its own command prompt.
- Disable automatic drive‐letter assignment to newly seen volumes by typing at the Diskpart command prompt:
automount disable - Get windows to remove any references to disks no longer in use (you will need to re-map any partitions)
automount scrub - Set the SAN policy so that all newly discovered disks except the boot disk will be offline and read-only by default (allegedly see tips below):
san policy= OfflineAll - Exit the Diskpart utility by typing:
exit - Check HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\MountMgr\NoAutoMount that it is set to 1
- Shut down Windows.
- Map the LUNs for the host in the Integrated Management Module of the SAN
- Boot the server
DiskPart command reference
- san policy= OfflineAll does not necessarily set the volumes to offline and read only!!!!
So you may need to do this manually
run diskpart
list disk
select disk x
attributes disk
attributes disk set readonly - If the disk is shown offline
list disk
select disk x
online disk
The disk should then show online and read only - do 1 and 2 for each SAN Disk
- Reboot server or restart the BE services to make sure Backup Exec sees the new drives
Tips
- Test this process first by creating a temporary LUN, creating a datastore on it and putting some sacrificial VMS on it to backup
- Make sure you read only the correct disks in Computer Management!
- In the Computer Management Console you can still attempt to delete the volume, but will get an error
- Make sure you have created the data store in VCenter on the LUN you are about to attach. Otherwise then you go into the Computer Management console, Windows will see it as unallocated and pop up a dialogue box to take control of it (cancel popup).
- If the disk is seen as unallocated, it won’t be available to Backup Exec to use as SAN transport. Restart the media server and so long as there is a datastore on that LUN, the disk should then appear in the Computer Management console as a basic disk and a Healthy, Primary Partition set as read only
- Basically there is conflicting “ best practices” out there 1.VMware recommends the SAN policy to be OnlineAll, but setting it to OfflineAll should guarantee the LUNS come into the computer management read only and offline. Then all you need to do is put them online.
- Windows server 2008 R2 does not do as it says on the tin 1.Setting the SAN policy to OfflineAll does not work properly. If you present one LUN at a time, it seems to put it online and read/write. If you present multiple LUNs quickly, it appears to make one online and read/write and the rest offline and readonly
Further Reading
http://webobjects.cdw.com/webobjects/media/pdf/symantec/Backup-Exec-2010-VMware-FAQ.pdf
http://www.symantec.com/business/support/index?page=content&id=TECH155831
Hi I found your article very interesting, is the first place I have seen that mention that the disks must be online.
I’m trying to do the same, but I still not able to do it, I’m running BE2014 my SAN is an EMC VNX5300 and vCenter 5.5, now I’m able to see all the LUNs on BE Server I just put the disks online as you said. but I still getting the error “The selected transport mode is unsupported by the current setup”
Any recommendation? I haven’t found any on BE nor VMware forums.
Regards
Hi,
Having to recharge the grey cells here as it was a while ago 🙂
Have you tried this one?
https://www.veritas.com/support/en_US/article.TECH210939
Whilst it says “certain machines” it is worth a look?
Yes I tried it, I did not find any on that folder. I think it is related to the VMware version since I did a test with an ESX running 5.5.0 update 1 and worked just fine, but the rest are 5.5.0 update 3 and so the vCenter. I just open a ticket with Veritas I hope they can help me.
Thank you for your answer.
Did you get an answer from Veritas regarding this?