Mailing List Archive: Cisco: NSP

simple link aggregation

Index | Next | Previous | View Threaded

mike-cisconsplist at tiedyenetworks

Jul 30, 2012, 9:24 PM

Post #1 of 7 (765 views)
Permalink

simple link aggregation

Hi,

I've googled and fiddled around and am unable to figure out simple link
aggregation. I have a 3650 on one side, a 2970 on the other, and two
100mbps links that I want to set up aggregation with to provide a
(virtual) 200mbps pipe between them. Yes I understand the load sharing
options and in this case I want to try round-robin. I don't seem to see
an easy/gentle how to do this and was hoping someone could tell me?

Mike-
_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

peter at rathlev

Jul 31, 2012, 1:06 AM

Post #2 of 7 (734 views)
Permalink

Re: simple link aggregation [In reply to]

On Mon, 2012-07-30 at 21:24 -0700, Mike wrote:
> I've googled and fiddled around and am unable to figure out simple
> link aggregation. I have a 3650 on one side, a 2970 on the other, and

3650? Or do you actually mean 3560, a switch? I'm assuming the latter.

> two 100mbps links that I want to set up aggregation with to provide a
> (virtual) 200mbps pipe between them. Yes I understand the load sharing
> options and in this case I want to try round-robin.

Cisco's hardware forwarding platforms only support deterministic hashed
load sharing. What link member a specific packet uses is determined from
L2, L3 or L4 information or a combination of these. The 3560 and 2970
platforms support using L2 and/or L3 information:

Switch(config)#port-channel load-balance ?
dst-ip Dst IP Addr
dst-mac Dst Mac Addr
src-dst-ip Src XOR Dst IP Addr
src-dst-mac Src XOR Dst Mac Addr
src-ip Src IP Addr
src-mac Src Mac Addr

Switch(config)#

This deterministic load sharing is much better than (per packet) round
robin, since it lets you avoid out-of-order packets. All packets to/from
the same source/destination uses the same link members and are thus
subjected to that same latencies et cetera.

> I don't seem to see an easy/gentle how to do this and was hoping
> someone could tell me?

This document describes etherchannels on Catalyst switches and is worth
a read:

http://www.cisco.com/en/US/tech/tk389/tk213/technologies_tech_note09186a0080094714.shtml

Configuring them is rather simple. Here's an example of an active LACP
port-channel trunk:

! *** Switch ***
default interface range FastEthernet0/1 - 2
interface range FastEthernet0/1 - 2
switchport
channel-group 1 mode active
!
interface Portchannel1
switchport trunk encapsulation dot1q
switchport mode trunk
switchport trunk allowed vlan 10-20
exit
!

Configuration for both CatOS and IOS can be found here:

http://www.cisco.com/en/US/tech/tk389/tk213/technologies_configuration_example09186a0080094470.shtml

--
Peter

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

mike-cisconsplist at tiedyenetworks

Jul 31, 2012, 9:34 AM

Post #3 of 7 (717 views)
Permalink

Re: simple link aggregation [In reply to]

On 07/31/2012 01:06 AM, Peter Rathlev wrote:
>
> Cisco's hardware forwarding platforms only support deterministic hashed
> load sharing. What link member a specific packet uses is determined from
> L2, L3 or L4 information or a combination of these. The 3560 and 2970
> platforms support using L2 and/or L3 information:
>
> Switch(config)#port-channel load-balance ?
> dst-ip Dst IP Addr
> dst-mac Dst Mac Addr
> src-dst-ip Src XOR Dst IP Addr
> src-dst-mac Src XOR Dst Mac Addr
> src-ip Src IP Addr
> src-mac Src Mac Addr
>
>

Damm, you are right. My problem is that I have many pppoe subscribers on
one side of the link, talking to essentially 1 server. So if rr isn't an
option, then I would likely need to do load-balance per source mac on
the subscriber side and by dest mac on the server side.

Thanks for the rest of the info, thats enough to get me started.

Mike-

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

diosbejgli at gmail

Jul 31, 2012, 11:49 AM

Post #4 of 7 (713 views)
Permalink

Re: simple link aggregation [In reply to]

You might want to use src-ip and dst-ip instead of mac depending on
your topology and on the location of the L2 boundary. If there's a
router in between, the src/dst mac might be static, being the router
mac.

Andras

On Tue, Jul 31, 2012 at 6:34 PM, Mike
<mike-cisconsplist [at] tiedyenetworks> wrote:
> On 07/31/2012 01:06 AM, Peter Rathlev wrote:
>>
>>
>> Cisco's hardware forwarding platforms only support deterministic hashed
>> load sharing. What link member a specific packet uses is determined from
>> L2, L3 or L4 information or a combination of these. The 3560 and 2970
>> platforms support using L2 and/or L3 information:
>>
>> Switch(config)#port-channel load-balance ?
>> dst-ip Dst IP Addr
>> dst-mac Dst Mac Addr
>> src-dst-ip Src XOR Dst IP Addr
>> src-dst-mac Src XOR Dst Mac Addr
>> src-ip Src IP Addr
>> src-mac Src Mac Addr
>>
>>
>
>
>
> Damm, you are right. My problem is that I have many pppoe subscribers on one
> side of the link, talking to essentially 1 server. So if rr isn't an option,
> then I would likely need to do load-balance per source mac on the subscriber
> side and by dest mac on the server side.
>
> Thanks for the rest of the info, thats enough to get me started.
>
> Mike-
>
>
> _______________________________________________
> cisco-nsp mailing list cisco-nsp [at] puck
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

peter at rathlev

Jul 31, 2012, 12:43 PM

Post #5 of 7 (709 views)
Permalink

Re: simple link aggregation [In reply to]

On Tue, 2012-07-31 at 09:34 -0700, Mike wrote:
> Damm, you are right. My problem is that I have many pppoe subscribers
> on one side of the link, talking to essentially 1 server. So if rr
> isn't an option, then I would likely need to do load-balance per
> source mac on the subscriber side and by dest mac on the server side.

It's always the sender doing the load-balancing, so that should be
possible by just using two different algos on each side of the link.
Alternatively, e.g. if you have more than just two switches, you might
get good results with "src-dst-mac".

I'm pretty sure none of the Catalyst switches can recognise PPPoE and
extract L3 information from the packets, so there's just the MAC
addresses.

--
Peter

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

nick at foobar

Jul 31, 2012, 1:24 PM

Post #6 of 7 (705 views)
Permalink

Re: simple link aggregation [In reply to]

On 31/07/2012 20:43, Peter Rathlev wrote:
> I'm pretty sure none of the Catalyst switches can recognise PPPoE and
> extract L3 information from the packets, so there's just the MAC
> addresses.

Nope they don't. src/dst ip address is as much as you can expect if it's
v4. If those headers aren't there, then it's src / dsc mac only. Not sure
if it supports ipv6 header hashing, but I suspect not - this sort of thing
would normally be explicitly documented if it were supported.

fwiw, other vendor switches have much better support for this than Cisco.

Nick

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

cconn at b2b2c

Aug 2, 2012, 11:52 AM

Post #7 of 7 (678 views)
Permalink

Re: simple link aggregation [In reply to]

> Not sure
>if it supports ipv6 header hashing, but I suspect not - this sort of thing
>would normally be explicitly documented if it were supported.

>fwiw, other vendor switches have much better support for this than Cisco.

Not sure if I am hijacking this thread, just curious if anyone has done src-dst-ip load balancing on a etherchannel with IPv6 on the 2970 as it seems to be possible;

show etherchannel load-balance
EtherChannel Load-Balancing Operational State (src-dst-ip):
Non-IP: Source XOR Destination MAC address
IPv4: Source XOR Destination IP address
IPv6: Source XOR Destination IP address

C.

_______________________________________________
cisco-nsp mailing list cisco-nsp [at] puck
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/

Index | Next | Previous | View Threaded

Interested in having your list archived? Contact Gossamer Threads