MPX 100's and switch redundancy question

[Jeff La]

I have a VMware environment with 3 host servers that each have 2 NICs dedicated to ISCSI traffic. My sorage is an EVA 6100 with dual mpx 100 controllers. I currently have 1-24 port Cisco 3750Gb switch (single point of failure) that the 6 ISCSI conections from the host servers as well as the 4 ISCSI ports on the 2 MPX controllers are attached to. I have just purchased another 3750 switch and plan to add it to my configuration as a stand alone switch, moving 1 ISCSI connection from each ESX server to it and also moving 1 connection from each mpx controller to provide redundancy. Is this the best approach to provide the ISCSI network redundancy or is there a better way? I could set the swtitches up in a stack configuration but don't see that as being very redundant.

[marsh_1]

looks pretty sound to me !

fwiw

[Pieter 't Hart]

you certaily will hav advantage by stacking the switches.

when stacking the 3750, they operate as a single switch not only for management but as a single switching backplane (through dedicated stack-cable).

in a 3750-stack you can create a load-balancing team across two switches!
in contrary to other switches that only form stack for management (utp or fiber port), you only have fault-tolerant team.

If both stackmembers have an uplink to the rest of the network you certainly have more redundancy, and if both switches are operational you have double active paths available.

[Jeff La]

Good point Pieter, but if one of the physical switch's ever has a problem or needs the firmware upgraded wouldn't the stack have to be taken down, which would mean no ISCSI traffic/no VM's? My thought was if they were stand alone switches one could be down for maintinance and you would still have a path to storage. I am trying to find the pro and cons of either method and appreciate any and all comments.

Thank you

[Pieter 't Hart]

Have not done FW upgrade on 3750 yet.

But afaik you can upgrade the FW and reload one at a time.
so allways one path is available.
I'll check if i can find some document on this.
( at this time only found "archive copy-sw" command in http://www.cisco.com/en/US/docs/switches/lan/catalyst3750/software/release/12.2_25_sea/command/reference/cli1.html )

[Pieter 't Hart]

There seems to be a single reload for the whole stack.
So if thats's a concern you should keep it as separate switches.

see witepaper http://www.cisco.com/en/US/prod/collateral/switches/ps5718/ps5023/prod_white_paper09186a00801b096a.html says :
========================
Cisco IOS Software Images Must Be Identical

The Cisco StackWise technology requires that all units in the stack run the same release of Cisco IOS Software. When the stack is first built, it is recommended that all of the stack members have the same level of software - either all SMI, all EMI, or AIPv6. This is because later upgrades of Cisco IOS Software mandate that all the switches to be upgraded to the same version as the master.
=======================
Upgrades Apply to All Devices in the Stack

Because the switch stack behaves like a single unit, upgrades apply universally to all members of the stack at once.
=======================

see upgrade procedure "Catalyst 3750 Software Upgrade in a Stack Configuration ...."
http://www.cisco.com/en/US/customer/products/hw/switches/ps5023/products_configuration_example09186a00804799d7.shtml
===============
Software images downloaded to the stack master are automatically downloaded to the rest of the stack members. Thus, you must be logged on to the master switch to perform this upgrade.
=================