The MD3000i is an active/active array meaning that both RAID controllers can service I/O.  However from the host's perspective to a single virtual disk (LUN) it us active/passave, meaning the one one RIAD controller owns a perticular virtual disk and will service all of it's I/O.  The other RAID controller can take ownership if there is a failure in the SAN preventing the host from accessing the virtual disk on the other RAID controller.  When configuring the MD3000i, it will load balance the virtual disk's ownership between the two RAID controllers.

 

You will also want to choose MRU policy for paths to the MD3000i.

Have you seen the Dell PowerVault Configuration Guide for VMware ESX/ESXi 3.5?

Jeff thanks for they reply - so MRU is the policy to use from the ESX to an MD3000i.

just another couple of questions.....

 

      Is there any way to map the paths (in my cas four) to the TCP sessions or controller pors??

ESX tells me which path is active 0,1,2 or 3 but how do I find out which IP address/pNic that is end-to-end

also

      How do I manually load ba;cae across the four paths.....

and

      Does each ESX server map the multi path to the same MD3000 in the same order i.e is path 0 on one ESX server the same subnet/raid ctrl port as another ESX server connected to the same MD3000?  

 

 

If you view the properties of Volume and select Manage Paths, you should see something like this:

vmhba32:0:0    iqn.1984-05.com.dell:power....   On

vmhba32:1:0    iqn.1984-05.com.dell:power....   On

vmhba32:2:0    iqn.1984-05.com.dell:power....   Active

vmhba32:3:0    iqn.1984-05.com.dell:power....   On

The number that I highlighted red, indicates the port on the MD3000i, the lowest number will be controller 0, port 0.  That number may not alway start with 0, depending if there are other iSCSI devices in the SAN that the ESX server is logged into.

In this case we can determine the following:

vmhba32:0:0    Controller 0, port 0    Standby

vmhba32:1:0    Controller 0, port 1    Standby

vmhba32:2:0    Controller 1, port 0    Active

vmhba32:3:0    Controller 1, port 1    Standby

 

The 0 at the end of the device, indicated the LUN that the virtual disk was assigned when it was mapped to the host with MDSM.

 

most likely if you have multiple LUNs assigned to the ESX server you will want to have the LUN balanced between the two RAID controllers.  You  can use MDSM and view/modify the perferred owner on the "Modify" tab.  I would recommend documenting the perferred owner for each virtual disk, then confirm the active path for each volume matches that.  You can change the Active path by disabling the paths you don't want to use until the path moves the the port you want it on, the be sure to re-enable all paths so that you have proper failover configured.

Hope this helps.

How many NICs are you using for iSCSI traffic?  How many subnets are you using for iSCSI traffic?

 

 

Jeff cheers for that info.

 

The iSCSI setup is two subnets vi two ethernet switches with two pNICs from each ESX server. Given that ESX doesn't load balance with software iSCSI initator didn't see the need to team multiple pNIC's, so it's just a question of looking at the traffic and moving the VD's about on the MD3000i (sigh) in an effort to get better backend I/O.

 

 

 

I running ESX 3.5 and have one Qlogic 4060 iSCSI HBA. When I check my manage paths on any LUN I only see one path. I set this up via Dynamic Discovery (Send Targets I think) and it automatically creates a single Static Discovery entry. I see all LUN's but only across one path and therefore one raid controller. I have one virtual disk with many smaller LUN's.

 

Am I limited to one path due to only having one HBA? Or am I misconfiguring somehting else? Please help! TIA

The MD3000i doesn't support Qlogic iSCSI HBAs.

 

On page 2 of the A14 support matrix (here <ADMIN NOTE: Broken link has been removed from this post by Dell>) it says "iSCSI offload is supported with the Broadcom NetXtreme II 5708c Network Adapter only".