[AusNOG] How to setup something like LACP across two switches

[Jason Leschnik]

Hey Mattia,

Faricpath on Cisco is TRILL based with Cisco extensions, SPB is the IEEE
competitor - Trill is IETF.

This would work between your different switches but the hosts still
wouldn't be able to form a single port-channel to those TRILL/FabricPath
enabled switches. Think of the Fabricpath core kinda like Layer3 core just
that we're routing Layer 2 frames as the payload. We still run into the
same limitation if we try to connect a host into two different switches, we
cannot form a single port-channel/LAG between them. So we're back to
needing vPC (MLAG/MCEC)

For this Cisco uses vPC+ which is vPC+FP to allow hosts to connect
Active-Active.

https://adamraffe.com/2013/03/04/vpc-and-vpc-understanding-the-differences/

On 20 April 2018 at 16:22, Mattia Rossi <mattia.rossi.mailinglists at gmail.com
> wrote:

> A bit late, I know, but I would have thought that IEEE 802.1aq (SPB) would
> solve this issue? (FabricPath on Cisco).
>
>
>
> On 20/04/2018 1:19 PM, James Cunningham wrote:
>
> Perfect thanks everyone for the replies - this is exactly what I was
> looking for.
>
> Spanning tree was also a quick, easy option too - but vPC was what I was
> looking for.
>
>
> On Fri, Apr 20, 2018 at 9:23 AM, Andrew Jones <Andrew.Jones at optus.com.au>
> wrote:
>
>> Here you go - https://www.cisco.com/c/en/us/
>> td/docs/switches/datacenter/nexus3000/sw/layer2/503_U2_1/b_
>> Cisco_n3k_layer2_config_guide_503_U2_1/b_Cisco_n3k_layer2_
>> config_gd_503_U2_1_chapter_01000.html
>>
>>
>>
>> Andrew Jones
>>
>> 0435 658 228
>>
>>
>>
>> *From:* AusNOG [mailto:ausnog-bounces at lists.ausnog.net] *On Behalf Of *James
>> Cunningham
>> *Sent:* Friday, 20 April 2018 9:18 AM
>> *To:* ausnog at lists.ausnog.net
>> *Subject:* [AusNOG] How to setup something like LACP across two switches
>>
>>
>>
>> Hello Fellow Ausnoggers,
>>
>>
>>
>> I'm hoping that I can quickly pick the brain of someone more
>> knowledgeable in data centre networking than myself.
>>
>>
>>
>> We have a customer in one of our racks in Equinix who has a single
>> network switch, and some servers connected to it. We currently have two
>> connections from our switch to the customer's switch, with LACP for
>> redundancy (and as a side effect, we get a slight bandwidth boost for 2 x
>> 1Gbps connections, which is a slight bonus).
>>
>>
>>
>> We would like to improve this by putting in two network switches on our
>> end, to protect again a single switch failure on our side, but the customer
>> will still have one single network switch.
>>
>>
>>
>> I'm pretty sure we can't do LACP with this style of setup - so what would
>> people recommend to achieve redundancy here? Main thing is that the
>> connection needs to auto-failover if a network switch fails, or if one of
>> the uplinks fails.
>>
>>
>>
>> I have created the attached diagram which illustrates what we are trying
>> to do.
>>
>>
>>
>> Can anyone please help? I'll owe a beer at the Next AusNOG meetup!
>>
>>
>>
>> Thanks
>>
>>
>>
>> James
>>
>
>
>
> _______________________________________________
> AusNOG mailing listAusNOG at lists.ausnog.nethttp://lists.ausnog.net/mailman/listinfo/ausnog
>
>
>
> _______________________________________________
> AusNOG mailing list
> AusNOG at lists.ausnog.net
> http://lists.ausnog.net/mailman/listinfo/ausnog
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ausnog.net/pipermail/ausnog/attachments/20180420/6c0e44d5/attachment.html>