VMware ESX iSCSI Set Up - Multi Pathing

Oh man! Ad-blocking software has been detected! :'(

This website is run by the community, for the community... and it needs advertisements in order to keep running. Blocking our ads means your killing our stats!
Please disable your ad-block, or become a premium member to hide all advertisements and this notice.

Wonder if anyone can look over the following as there is not too much in the way of documentation from VMware on this particular config.

I will try to explain it best I can so please bare with me.

We have the following equipment:

iSCSI SAN with 16 disks that run Windows Storage Storage. Since there is just one motherboard I will take that to equal one controller. There are 4 iSCSI ports on the back that connect to dedicated iSCSI switches on the back of a blade center. Config is as follows:

Port1 10.10.1.1 Connects to a port on switch 1 of the BC
Port2 10.10.2.1 Connects to a port on switch 1 of the BC
Port3 10.10.3.1 Connects to a port on switch 2 of the BC
Port4 10.10.4.1 Connects to a port on switch 2 of the BC

The ESX hosts are blades with 2 NICS for iSCSI. One NIC connects to switch 1 and the other to switch 2:

NIC1 to switch 1
NIC2 to switch 2

I have followed the iSCSI config guide and not teamed these NICS to a single vmkernel port group. Following the VMware the suggestion is a 1:1 mapping of each VMkernel to a NIC on the vSwitch. However if I were to do this then I would only be able to assign one address to each NIC and since I have only two on each host I will only be able to use say for example 10.10.1.2 and 10.10.2.2.

So...instead I have created 4 vmk port groups and mapped 2 to each NIC like so:

vmk0 10.10.1.2 Mapped to NIC1
vmk1 10.10.2.2 Mapped to NIC1
vmk2 10.10.3.2 Mapped to NIC2
vmk3 10.10.4.2 Mapped to NIC2

So far it seems to work just fine, each LUN has 4 paths.....however I would like any input from anyone who has encountered a similiar thing before.

The main concern is path selection. I have tried "fixed" and this seems to bind one path (1GB) to the LUN and the load balancing seems to consist of having one different path to each LUN. I have also tried round robin which again just worked fine however I was under the impression for storage to avoid this due to path/LUN thrashing, although with a single controller this may not be an issue.

Any input most welcomed......Boy do I prefer FC!!!

 

Oh man! Ad-blocking software has been detected! :'(

This website is run by the community, for the community... and it needs advertisements in order to keep running. Blocking our ads means your killing our stats!
Please disable your ad-block, or become a premium member to hide all advertisements and this notice.

Just to confirm, the IP's given have a mask of 255.255.255.0 so each are on a different subnet as per the SAN provider.

Dynamic discovery is used on all four addresses and it is configured by IP of the initiator on the SAN side not by IQN.