Hi Yubao,
Thanks for the comment. I think more and have not understood too how
"mass-withdraw" is possible with single-homed sites.
Section 5 instructs us to use ESI=0 for single-homed sites.
EVI could have many Ethernet Segments connected on 1 PE, all would be with
ESI=0 for single-homed.
Then I do not see what to signal to the remote to start "mass-withdraw".
{EVI=X, ESI=0} would mean many physical Ethernet Segments in this case, no
possibility to point to one particular Ethernet Segment (with broken port to
CE).
Hence, if one would like to have "mass-withdraw"
Then all Ethernet Segments should be numbered (ESI>0).
It is probably a little violation of RFC 7432 because ESI should be 0 for
single-homed (section 5).
But it would probably work because it would look like a situation with a
redundant ESI connection permanently down.
I am not 100% sure that would be any other negative consequences.
Eduard
From: wang.yub...@zte.com.cn [mailto:wang.yub...@zte.com.cn]
Sent: Thursday, December 9, 2021 4:37 AM
To: Vasilenko Eduard <vasilenko.edu...@huawei.com>
Cc: yuya...@yuyarin.net; bess@ietf.org
Subject: Re: [bess] Contradiction for the RFC 7432 definition of the fast
convergence (withdrawal) for single-homed CEs
Hi Eduard,
I don't think an Ethernet A-D per ES route with a zero ESI is better to use,
Because each single-homed CE is an individual ES,
that's why MAC mobility happens between two zero ESIs (for different
single-homed ethernet segments)
but not happens between the same ESI (local and remote).
And I think such Ethernet A-D per ES route can't achieve the expected
"mass-withdraw" behavior actually,
Because it may not influence the installing of the RT-2 routes whose ESI are
zero.
If we want to use such Ethernet A-D per ES route to achieve mass-withdraw
behavior,
I think it will be difficult for us to decide when to trigger the mass-withdraw
too.
Please see section 9.2.2 of RFC7432:
"9.2.2. Route Resolution
If the Ethernet Segment Identifier field in a received MAC/IP
Advertisement route is set to the reserved ESI value of 0 or MAX-ESI,
then if the receiving PE decides to install forwarding state for the
associated MAC address, it MUST be based on the MAC/IP Advertisement
route alone."
Regards,
Yubao
On Wed, 8 Dec 2021 09:28:27 +0000
Vasilenko Eduard
<vasilenko.edu...@huawei.com<mailto:vasilenko.edu...@huawei.com>> wrote:
> Hi Yuya,
> Thanks.
> Your explanation looks reasonable because section 8.2:
> " If no other PE had advertised an Ethernet A-D route for the same segment,
> then the PE that received the withdrawal simply invalidates the MAC entries
> for that segment."
> Does not say "MUST" or "SHOULD".
> But the word "may" in this sentence was better to use.
> OK. I understood. It is an optional feature that was not claimed optional
> plainly.
> Eduard
> -----Original Message-----
> From: BESS [mailto:bess-boun...@ietf.org] On Behalf Of Yuya KAWAKAMI
> Sent: Wednesday, December 8, 2021 3:58 AM
> To: Vasilenko Eduard
> <vasilenko.eduard=40huawei....@dmarc.ietf.org<mailto:vasilenko.eduard=40huawei....@dmarc.ietf.org>>;
> bess@ietf.org<mailto:bess@ietf.org>
> Subject: Re: [bess] Contradiction for the RFC 7432 definition of the fast
> convergence (withdrawal) for single-homed CEs
>
> Hi Eduard,
>
> As my understanding, these statements are not contradictory because mass
> withdrawal is an optional functionality.
> Section 8.3 says when PEs are operating All-Active redundancy mode, Ethernet
> A-D per Ethernet Segment Route must be advertised for split horizon.
> This would be reason why section 8.2.1 says "not needed" in case of
> single-homed scenarios.
>
> I understand these sentences as:
> - Implementations can use Ethernet A-D per ES routes to achieve mass
> withdrawal for single-homed CE (optional)
> - If Implementations want to achieve mass withdrawal, Ethernet A-D per ES
> routes should be used
>
> The implementation I'm using does not support mass withdrawal for
> single-homed CE.
>
> If there is any misunderstanding, I would appreciate it if you could point it
> out.
>
> Thanks,
> Yuya
>
> On 2021/12/08 4:24, Vasilenko Eduard wrote:
> > Hi EVPN guru,
> >
> > It looks like RFC 7432 section 8.2.1 (Constructing Ethernet A-D per
> > Ethernet Segment Route) has an error:
> > "The Ethernet A-D route is not needed when the Segment Identifier is set to
> > 0 (e.g., single-homed scenarios)."
> >
> > Because without "per ES route" it would not be possible to signal
> > "mass withdrawal" If CE-PE connection would fail That plainly promised for
> > single-homed CEs in section 8.2:
> > " If no other PE had advertised an Ethernet A-D route for the same segment,
> > then the PE that received the withdrawal simply invalidates the MAC entries
> > for that segment."
> > Or implied in section 17.3:
> > "The Ethernet A-D per ES routes should be used by an implementation to
> > optimize the withdrawal of MAC/IP Advertisement routes."
> >
> > Have I missed something?
> >
> > Eduard
> >
> > _______________________________________________
> > BESS mailing list
> > BESS@ietf.org<mailto:BESS@ietf.org>
> > https://www.ietf.org/mailman/listinfo/bess
> >
>
> _______________________________________________
> BESS mailing list
> BESS@ietf.org<mailto:BESS@ietf.org>
> https://www.ietf.org/mailman/listinfo/bess
>
>
>
_______________________________________________
BESS mailing list
BESS@ietf.org
https://www.ietf.org/mailman/listinfo/bess
