This most definitely works, with supported switches. I'm using a pair of 3750Es, with a link from each ESX host on each switch, and two links from each filer. The ESX links are configured "on" and the filer links with LACP.
There is more detail in TR3428 (http://www.netapp.com/us/library/technical-reports/tr-3428.html) and in VMWorld session TA2784 (which you get from the VMWorld site or there might be a copy out there if you Google hard enough).
Cisco isn't the only vendor with this kind of functionality, but they do seem the most prevalent.
Enjoy!
Peter
-----Original Message----- From: David Lee [mailto:t.d.lee@durham.ac.uk] Sent: Wednesday, February 04, 2009 7:27 AM To: toasters@mathworks.com Subject: MEC (Multichassis EtherChannel)
(I'm a little out of my depth here, so there may be some bits needing lateral interpretation...)
We already use bonded/trunked ethernet from the NetApp to a switch. That's fine.
We are planning a network upgrade, which gives us a chance to do this bonding/trunking to a logically linked pair of switches (Cisco 3750E) that operate as a single virtual switch, using "MEC". (The theory is that if a switch in the MEC-pair fails, the bond/trunk link continues working through the other switch, albeit with degraded throughput.)
Does anyone know whether MEC does/doesn't (will/won't) work? Is there anything MEC-specific that needs to be done in NetApp? Issues? (Other questions I should be asking? Etc.)
Pointers to existing information would probably be fine.
Thanks in advance.