This is my ISCSI setup.
Storage:
EMC NX4 2 blades Active/Standby with (4) 1GIG links cge0-cge3 each
cge0-1 setup as lacp with ip of 10.1.1.1/24 on switch 1 VLAN 10
cge2-3 setup as lacp with ip of 10.1.2.1/24 on switch 2 VLAN 20
I did this for redundancy at the physical switch and EMC link level on seperate.
Host Setup has 2 physical nics for ISCSI (software adapters - vmhva37) and im running ESXi 4.1
VSwitch 3
VMKernel port ISCSI1
VMK: 10.1.1.10/24 - Switch 1 ( separate subnet from ISCI2 and not routed)
VSwitch 4
VMKernel port ISCSI2
VMK: 10.1.2.10/24 -Switch 2 ( separate subnet from ISCSI1 and not routed)
So i have 1 pnic one each iscsi network going to to each switch, in turn going to each EMC lacp. This should provide for redundancy at the pnic level, switch level and emc link level.
Questions...
1. Should i have both host iscsi nics on 1 vswitch or is my multiple iscsi vswitch setup ok?
2. If i keep multiple iscsi vswitches with a nic in each, do i still need to bind them as if you would if they were both in a single vswitch?
esxcli swiscsi nic list -d vmhba37(iscsi software adapter) doesn't show me anything, is this normal for a multiple iscsi vswitch? Or is binding still required even in a multi vswitch setup?
3. What should my throughput look like and how can/should i test it?
4. how many luns per target or should it be a 1-1 ratio?
5. Path selection is set to Fixed by default, should i leave this as is?
As is, i have 2 test luns(20 and 21). On my storage adapters page i see 2 paths for each lun
vmhba37;C0;T0;L20 - 10.1.1.1 active(I/O)
vmhba37;C0;T0;L21 - 10.1.1.1 active(I/O)
vmhba37;C1;T0;L20 - 10.1.1.2 active
vmhba37;C1;T0;L21 - 10.1.1.2 active
It took me about 30-40 minutes to copy a 50Gig file from the local host storage to one of my test LUNS. Excessive?
How does all of this look, am i on the right path? Any thoughts, recommendations?
I appreciate any assistance and guidance.
Nas
