VMware Cloud Community
cliess
Enthusiast
Enthusiast

IBM DS4800 vSphere 4/ESXi LAN-Free Backup Question

Hi Everyone -

We're in the final stages of planning our ESX 4.0.0 Update 1 -> ESXi

4.1 upgrade and as part of this, we're going to move to LAN-free

backups via the vStorage API. We've already tested this with VMs on our iSCSI SAN and it

works like a champ. Last up is VMs that live on our IBM DS4800.

For our test, we zoned in the HBA of our backup server (Running Windows

2003 Standard R2 w/ SP2 + vRanger Pro DPP v4.5.3) and created the host

and port identifier in Storage Manager (v10.50). At this point, we were

hoping to simply allow access to one of the production LUNs with some

low-impact VMs on the odd chance that presenting the disk to both a

Windows + ESX server would cause some kind of corruption.

Unfortunately, since our ESX hosts all live within a Host Group, it

appears as though the only method of testing would be to add our test

backup host into our ESX server Host Group, which naturally has access

to all of our VMFS LUNs.. Is this accurate?

Are we going about doing this the right way? Any help would be very much appreciated.

Thanks!

-Craig

Reply
0 Kudos
7 Replies
cliess
Enthusiast
Enthusiast

Anyone? Any help would be greatly appreciated!

-Craig

Reply
0 Kudos
ThompsG
Virtuoso
Virtuoso

Hi there Craig,

Not sure I'm going to be much help but will give it a go Smiley Wink No warranty or responsibility implied.

Looking at the IBM DS4800 it appears to have similar concepts to the EMC CLARiiONs so wondering if you could do something like this:

- Create another Host Group called vcb_proxy

- Put your VCB server in here

- Assign the LUN to be backed up to this Host Group as well as to the ESX Host Group

On the CLARiiON, or rather NaviSphere, you will get a warning about LUNs assigned to multiple Storage Groups but it does allow you to continue. The concern is centered around different OS/hosts access the same LUNs - so can be ignored in this case. Well not ignored exactly but duly noted Smiley Happy

If it was me I would try with an empty LUN (VMFS formatted, etc) just in case it removes it from the ESX Host Group and adds to the other. As an extra belts and bracers approach I would even go to creating a test ESX host goup, i.e. host group with test ESX server, 1 LUN, host group with VCB server accessing the same LUN - but this may just be overkill. Over to you on this one.

Sorry if this is obvious but make sure you have automount disabled, etc on the vcb proxy before begining.

Kind regards.

idle-jam
Immortal
Immortal

vRanger DPP does not use VCB anymore but VADP. The only difference is that you will not need to install the vcb software as VADP came inbuilt with VADP.

On top of the above advise, you can go a mile further and only assign "read only" access for the newly created host group. this will ensure that the presented LUN to the vranger server gets wipe out automatically via automount. You will need to disable automount first if you do not do read only.

On vRanger box, From the start menu, select “run” and enter diskpart.

Run the automount disable command to disable automatic drive letter assignment.

Run the automount scrub command to clean any registry entries pertaining to previously mounted volumes


iDLE-jAM | VCP 2, VCP 3 & VCP 4

If you found this or any other answer useful please consider the use of the Helpful or correct buttons to award points

cliess
Enthusiast
Enthusiast

Hi -

This is exactly the route we are going. Unfortunately, it doesn't appear like you can assign LUNs to more than one host group, so the plan is to evacuate an entire LUN using Storage vMotion, remove one host from the cluster, create a new host group with just that one host + the vRanger server + the newly-evacuated LUN and test.

Will let you know how it goes. Thanks for the reply!

-Craig

Reply
0 Kudos
cliess
Enthusiast
Enthusiast

Just wanted to follow up with what we ultimately did for the test:

  1. Put one of the hosts into maintenance mode

  2. Removed it from the cluster

  3. Storage vmotioned all VMs off one of the LUNs

  4. In Storage Manager, we created a new Host Group and moved in the test vRanger & ESX host, as well as the newly-evacuated LUN.

  5. Cabled in the test vRanger host to the SAN. Used an old 2GB HBA for this test.

  6. Restored a VM from the test vRanger host

  7. Created a new backup job for the newly-restored VM and selected to use the Fiber/iSCSI backup option.

  8. The backup ran and didn't blow anything up. Horray!

  9. Undo everything above to get the host back into the original cluster.

Performance wasn't very good, but our test vRanger host is equally as poor, so we are attributing the slower-than-expected performance numbers to that. Also, we didn't bother to install the IBM DSM, so there's a chance that was holding us back as well.

Thanks to the few of you who chimed in to assist! Smiley Happy

-Craig

Reply
0 Kudos
idle-jam
Immortal
Immortal

How much MB/sec are you getting with vranger?


iDLE-jAM | VCP 2, VCP 3 & VCP 4

If you found this or any other answer useful please consider the use of the Helpful or correct buttons to award points

Reply
0 Kudos
cliess
Enthusiast
Enthusiast

We're seeing big fluxuations in LAN backups.. anywhere betwen 8-30MB/s. This is still much faster than we got with vRanger 3, though.

Will be moving to iSCSI-based backups shortly......

-Craig

Reply
0 Kudos