cancel
Showing results for 
Search instead for 
Did you mean: 

Netbackup Appliance VMWare configuration / Zoning

Kevin_O_Connor
Level 4

Hi All,

 

I'm just looking for some clarification on configuring VMWare backups using the agent.

 

My master is running 7.6.1.1 as are our 5230 Appliances and vSphere is 5.5 so support is not an issue.

I've set aside two ports on each appliance for vmware backups so I know where I want to zone to on that end but what else do I zone to ?

Do I need to zone in the ESX servers or just the WWN's of the storage arrays ?

 

The VMWare admin guide states the following but isn't very clear.

"To use the SAN transport type, set up the datastore on Fibre Channel or iSCSI. In this configuration, the VMware backup host must be

able to access the datastore over the SAN."

The ESX boxes are already san connected.

We tested the agent a few years ago to VTL with a Windows Off-Host server and found it to be slower than just sending it across the LAN but need and want to test it again especially with the 5230's now bedded in.

 

Thanks in advance.

2 ACCEPTED SOLUTIONS

Accepted Solutions

RiaanBadenhorst
Moderator
Moderator
Partner    VIP    Accredited Certified

Do I need to zone in the ESX servers or just the WWN's of the storage arrays ?

You only need to zone the storage arrays to the Appliance so it can access the luns (as the ESX does).

View solution in original post

sdo
Moderator
Moderator
Partner    VIP    Certified

You should use single initiator zoning - this is best practice.

Think of the appliance as an ESXi server, in that it reads (backs-up) and writes (restores) - so I/O is essentially the same as an ESXi host.

So, zoning would be like:

$ alicreate AL_APPLIANCE1_HBA1_PORT1, "nn:nn:nn:nn:nn:nn:nn:nn"

$ alicreate AL_ESX1_HBA1_PORT1, "nn:nn:nn:nn:nn:nn:nn:nn"

$ alicreate AL_STORAGE1_FE1_PORT1, "nn:nn:nn:nn:nn:nn:nn:nn"

$ zonecreate ZN_APPLIANCE1_HBA1_PORT1, "AL_APPLIANCE1_HBA1_PORT1; AL_STORAGE1_FE1_PORT1"

$ zonecreate ZN_ESX1_HBA1_PORT1, "AL_ESX1_HBA1_PORT1; AL_STORAGE1_FE1_PORT1"

$ cfgadd CFG_FAB_A, "ZN_APPLIANCE1_HBA1_PORT1; ZN_ESX1_HBA1_PORT1"

$ cfgsave

$ cfgenable CFG_FAB_A

And if you present a VMFS data store LUN to both server WWPNs, i.e. of APPLIANCE1 and ESX1, then both the appliance and the ESX host will see the same LUN with the same LUN number (i.e. avoid using LUN masking - to make your life easier) - and you should be good to go.

Do you see how you need a minimum of three aliases - one for each node, and one for the storage array.  In practice you'll have many many more - for multi-pathing, and fabric resiliency?

Post back if you'd like a larger more comprehensive multi-path'd multi-fabric zoning example.

HTH.

View solution in original post

11 REPLIES 11

RiaanBadenhorst
Moderator
Moderator
Partner    VIP    Accredited Certified

Do I need to zone in the ESX servers or just the WWN's of the storage arrays ?

You only need to zone the storage arrays to the Appliance so it can access the luns (as the ESX does).

sdo
Moderator
Moderator
Partner    VIP    Certified

You should use single initiator zoning - this is best practice.

Think of the appliance as an ESXi server, in that it reads (backs-up) and writes (restores) - so I/O is essentially the same as an ESXi host.

So, zoning would be like:

$ alicreate AL_APPLIANCE1_HBA1_PORT1, "nn:nn:nn:nn:nn:nn:nn:nn"

$ alicreate AL_ESX1_HBA1_PORT1, "nn:nn:nn:nn:nn:nn:nn:nn"

$ alicreate AL_STORAGE1_FE1_PORT1, "nn:nn:nn:nn:nn:nn:nn:nn"

$ zonecreate ZN_APPLIANCE1_HBA1_PORT1, "AL_APPLIANCE1_HBA1_PORT1; AL_STORAGE1_FE1_PORT1"

$ zonecreate ZN_ESX1_HBA1_PORT1, "AL_ESX1_HBA1_PORT1; AL_STORAGE1_FE1_PORT1"

$ cfgadd CFG_FAB_A, "ZN_APPLIANCE1_HBA1_PORT1; ZN_ESX1_HBA1_PORT1"

$ cfgsave

$ cfgenable CFG_FAB_A

And if you present a VMFS data store LUN to both server WWPNs, i.e. of APPLIANCE1 and ESX1, then both the appliance and the ESX host will see the same LUN with the same LUN number (i.e. avoid using LUN masking - to make your life easier) - and you should be good to go.

Do you see how you need a minimum of three aliases - one for each node, and one for the storage array.  In practice you'll have many many more - for multi-pathing, and fabric resiliency?

Post back if you'd like a larger more comprehensive multi-path'd multi-fabric zoning example.

HTH.

sdo
Moderator
Moderator
Partner    VIP    Certified

You don't zone the Appliances and ESXi hosts togther.

You need to zone the appliance just like as if it were an ESX host.

stracy19
Level 2
Partner

We have 5230, HP EVA 6500 and ESX with 5.5   We have single WWN zoning that has the 5230 and the HP EVA array  (so one zone with 5230 and Disk storage) 

We see the devices in the appliance on a FC scan once they are presented from the HP EVA.   The EVA is active/active

We get a 23/6 error complaning about the LUN's not being seen

This backup works fine in NBD mode, se we know the accoutn is good.  We also see the snap shotsbeing taen in vCentre. 

 

I have heard you don't zone the ESX in with the appliance and storage but in your above exampe you have all three together.   

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified
Have you also done lun assignment for the Appliance on HP EVA management software?

RiaanBadenhorst
Moderator
Moderator
Partner    VIP    Accredited Certified

Create the vxms logs to see what is happening.

 

https://support.symantec.com/en_US/article.HOWTO70905.html

sdo
Moderator
Moderator
Partner    VIP    Certified

Hi Kevin,

You said "I have heard you don't zone the ESX in with the appliance and storage but in your above exampe you have all three together."...

...if you look at the example again, yes you'll see three aliases, but only two zones, each single inititator.  i.e. the first zone is "ESX initiator; storage target" and the second zone is "Appliance initiator; storage target".  Do you see how the ESX host is zoned to storage, and the appliance is zoned to storage.

I assumed that you already have the ESX hosts zoned, so... in zoning terms, just pretend that the appliance is another ESX host.  We ourselves know that the appliance is not an ESX host, but the fabric doesn't know that... the fabric just sees the appliance as yet another server (initiator) configured (zoned) to access storage (target).

HTH.

stracy19
Level 2
Partner

Come to find out the cusotmer was getting us to use LUN's that had not SAN VM's on them. We had it setup correctly in the end. 

sdo
Moderator
Moderator
Partner    VIP    Certified

Glad you got it working.  Did we help?  If so, mark any posts that helped you as solution?

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified
Seems someone else has hijacked this discussion. Kevin O' has not been back since he posted...

Kevin_O_Connor
Level 4

Sorry Folks,

 

Between EV/Netbackup upgrades and H/W refreshes across our sites I haven't had a minute to get back to this.

 

I've had to put this testing on hold especially as we've now got connectivity issues to the new 3Par's where our virtual servers are now going to be hosted.

 

In saying that though I'm confident with the solution marked above (Thanks Marianne) I'll be able to get this configured ad start testing soon enough.