[72846] in North American Network Operators' Group
Re: COnfiguration Suggestion - Etherchannel
daemon@ATHENA.MIT.EDU (Mike Sawicki)
Tue Jul 27 22:51:39 2004
Date: Tue, 27 Jul 2004 22:45:10 -0400
From: Mike Sawicki <fifi@HAX.ORG>
To: "Richard J. Sears" <rsears@adnc.com>
Cc: Nanog <nanog@nanog.org>
In-Reply-To: <20040727192308.D925.RSEARS@adnc.com>
Errors-To: owner-nanog-outgoing@merit.edu
On Tue, Jul 27, 2004 at 07:23:29PM -0700, Richard J. Sears wrote:
>
> Hey Everyone,
>
> I am building out a customer that needs more than 1000Mbps of sustained
> bandwidth. Because of the customer equipment, etherchannel was suggested
> as the means to do this (it is compatible with this customers equipment).
>
> I am running a 6509 with Dual SUP720's in IOS mode only (no cat software).
>
> It was pointed out that there are really two different ways to configure
> the switch - I guess my question is which is the best (lowest overhead,
> etc)? Hopefully someone out there has been down this road before.
>
> TIA
>
> Two methods:
>
> !
> interface Port-channel2
> no ip address
> switchport
> switchport access vlan 10
> switchport mode access
> !
> interface GigabitEthernet7/1
> no ip address
> switchport
> switchport access vlan 10
> switchport mode access
> channel-group 2 mode on
> !
> interface GigabitEthernet8/1
> no ip address
> switchport
> switchport access vlan 10
> switchport mode access
> channel-group 2 mode on
> !
> interface Vlan10
> description Customer_Name
> ip address 192.168.0.1 255.255.0.0
> no ip redirects
> no ip unreachables
> !
>
>
>
> And then there is this way:
>
>
>
> !
> interface Port-channel2
> description Customer_Name
> ip address 192.168.0.1 255.255.0.0
> no ip redirects
> no ip unreachables
> !
> interface GigabitEthernet7/1
> description Customer_Name EtherChannel Interface #1
> no ip address
> channel-group 2 mode on
> !
> interface GigabitEthernet8/1
> description Customer_Name EtherChannel Interface #2
> no ip address
> channel-group 2 mode on
> !
>
I would use method #2 above.. L3 FEC produces better balancing
results as it is flow based, rather than mac-based. I'm not 100%
certain that using the SVI interface would not produce a proper
balance, but I doubt it. Using method one I would expect only one
of the links to be used.
Use method 2 especially if you mean this to be a L3 handoff to the
customer.
Mike Sawicki (fifi@HAX.ORG)