Switch Port Configuration for Highly Available Servers

[mattrmurdock]

     Is there any special setup that is recommended when a Highly Available Cluster solution is attached to Data Center Cisco Switches or is default settings for the switch ports sufficient? I do have each NIC split across seperate switches/blades (or at least try to communicate that).

     On the OS side we will use either Bonding under Linux; IPMP or Aggregated Links under Solaris; APA (aggregated ports) under HP/UX. I do not always have the opportunity to discuss this in detail with my customers networking team, since they barely understand English (Most are in the APAC countries).

     I understand its not a simple question to answer. There is alot of detail that isnt being addressed, but I just need a general guideline to give to my customers if one is available.

[Reza Sharifi]

For higher availability and redundancy you may want to use Etherchannel if you are not already using it.

HTH

Sent from Cisco Technical Support iPad App

[Peter Koltl]

As long as the NIC teaming (under any above term) is configured active-standby (i. e. only one card sending traffic) you don't even need to configure etherchannel on the switch. Just make sure both ports are in the same VLAN. I'm not sure whether the standby NIC accepts incoming frames that the switch sends if you enable etherchannel.

As long as the NIC teaming is configured without LACP (i.  e. no conversation between NIC and switch about which ports are aggregate members) you don't need to (and shouldn't) configure  LACP on the etherchannel on the switch.

If both sides talk LACP, etherchannel will be happy.

[Maher Abdelshkour]

Hello Matthew,

You can have full info about this here:

http://goo.gl/1Gj5Lo