Re: draft-mmm-bfd-on-lags: IPv4 _and_ IPv6 ?

[Marc Binderberger <[e-mail]>]

Hello Jeff,

> The implication is that if IPv4 drops, but IPv6 stays up, that the LMM
> should remove *only* the impacted address family from use.


... if the implementation can differentiate between IPv4 and IPv6 in the load balance. I have no intention to make such an AF-aware load-balance table a requirement. The fact that either an IPv4 or an IPv6 BFD failure can remove the link for all traffic is understood by (my) customers and acceptable.

But with such an "if an implementation's distribution algorithm can differentiate between IPv4 and IPv6 then ..." condition I can add something to the draft.


Has anyone implemented such an AF-ware load-balancer?  Just for my curiosity.


Regards, Marc



On 2012-09-12, at 15:34 , Jeffrey Haas wrote:

> Just to prod this particular topic a bit:
> 
> On Thu, Aug 16, 2012 at 05:20:04PM +0200, Marc Binderberger wrote:
>> Hello Manav, Pablo et al.,
>> 
>> I think we will work on an update with a better wording. What we have now can be misunderstood. I agree with Manav's comment that what we really want to enforce is:
>> 
>> when you run a micro session for a particular address family on one member link then you must run micro sessions for this address family on all member links (of the LAG).
>> 
>> 
>>> Do you have some scenarios in mind?
>> 
>> I have a (big) customer in mind who was just asking for this :-)
>> The reason is that BFD micro sessions do cover layer-3 aspects as well and the customer thinks we should cover both, IPv4 and IPv6 code path. Sounds reasonable.
> 
> Let us presume that we arrive at wording that effectively says "if you run
> micro sessions in a given address family, do it on all links".
> 
> Presume also you run both IPv4 and IPv6.
> 
> The implication is that if IPv4 drops, but IPv6 stays up, that the LMM
> should remove *only* the impacted address family from use.
> 
> However, we're also suggesting that a scenario where one runs only one
> address family across the LAG is sufficient to impact all layer 3 traffic
> for the LMM.  Harkening back to the IS-IS observations, I'm fine with this.
> 
> What this means is we need language that says all of this.
> 
> -- Jeff (speaking as an individual contributor)
> 

--
Marc Binderberger           <[e-mail]>