VMware Cloud Community
snootalope
Contributor
Contributor
Jump to solution

Network config for SAN traffic

Please see the attached image. One is what I currently have, the after one is what I'm thinking I need to do.

Here's my issue; I think I've setup the networking incorrectly for my esx boxes to hit the SAN. At the moment, it's all working just fine, however I think I may have done some extra streps that I didn't need to do. Right now, I've got a NIC deticated to two different vSwitches that are used for nothing but SAN traffic. the more I look at this, the more I think I should have used one single vSwitch on each esx box and just added the needed console and kernel ports there. So, I'd have one single vswitch on each box dedicated to SAN traffic, each vSwitch would have the two current NICs with two console and two kernel ports.

Maybe I'm just beating a dead horse here since they'll both accomplish the same thing, yet maybe it's a smarter and simpler move to go with the after scenario since they can load balance between the two and failover would work better. ??? My biggest concern is, can I had two different console and two different kernel ports, both on different subnets, to the same vSwitch?

Thanks for any advice.

0 Kudos
1 Solution

Accepted Solutions
BKarciauskas
Enthusiast
Enthusiast
Jump to solution

Yes your thinking is right. having the two NIC's on a single vswitch will give you more options for failover and load balancing. Also having multiple console ports and kernal ports on the same vswitch isn't an issue.

Ben Karciauskas

Cheers, Ben Karciauskas Blog - http://karciauskas.wordpress.com

View solution in original post

0 Kudos
6 Replies
BKarciauskas
Enthusiast
Enthusiast
Jump to solution

Yes your thinking is right. having the two NIC's on a single vswitch will give you more options for failover and load balancing. Also having multiple console ports and kernal ports on the same vswitch isn't an issue.

Ben Karciauskas

Cheers, Ben Karciauskas Blog - http://karciauskas.wordpress.com
0 Kudos
RParker
Immortal
Immortal
Jump to solution

I've got a NIC deticated to two different vSwitches that are used for nothing but SAN traffic.

That's not possible, if you assigned a NIC to a switch, you can't assign that same nic to another switch at the same time. And you can always delete a switch, just don't delete the switch that's managing your ESX host (connected to vCenter)

0 Kudos
snootalope
Contributor
Contributor
Jump to solution

Sorry, typo on my part, that should have read "...... two different NICs deticated to tow different vSwitches...."

0 Kudos
snootalope
Contributor
Contributor
Jump to solution

Ok.. I'm missing something here. Our SAN has dual storage processor's each has two NIC's configured like so:

SAN

SPA #1 - 192.168.20.4.............Runs to SAN Switch#1

SPA #2 - 192.168.21.4.............Runs to SAN Switch#2

SPB#1 - 192.168.20.5..............Runs to SAN Switch#1

SPB#2 - 192.168.21.5..............Runs to SAN Switch#2

The two switches are dedicated to this san traffic only.

In each of my esx hosts I have two NIC's. Each nic runs to a different switch. Is this not enough to provide failover to the SAN? If I add all those addresses to my storage adapters dynamic list, it works just fine and I can vmkping all four address from my host. But if I unplug one of the NIC cables from the ESX host, I can't ping all the address above, it'll only let me ping one or the other of the subnets.

I'm looking for failover here, but I can't figure out where I'm going wrong...

0 Kudos
BKarciauskas
Enthusiast
Enthusiast
Jump to solution

Is there a reason that you have put them on different subnets? Since you have dedicated switching for the storage network there probebly isn't a reason to segregate traffic.

I would put all IP's on the same subnet and test. Let me know how you go.

Cheers, Ben Karciauskas Blog - http://karciauskas.wordpress.com
0 Kudos
snootalope
Contributor
Contributor
Jump to solution

Good question.. I originally set this up with everything san on the same subnet, but that had some negative results on the esx boxes. it'd take 10 minutes to rescan the iscsi targets. Reading through the EMC guide, it says to use two different subnets......so that's what I went with. It'd be an easy fix I think if I didn't have to worry about two different ones though.

0 Kudos