Re: [bess] FW: WG Last Call (including implem status & shepherd) for draft-ietf-bess-evpn-vpws-03

["Jeffrey (Zhaohui) Zhang" <[e-mail]>]

Hi Jorge, Ali,

> [JORGE2] In VPWS DF election is needed for single-active. The NDF will
> block the AC for the service.

> [JORGE] This is single active, so only one can transmit/receive to/from
> the CE. So only that one should send P=1.

If I understand it correctly, the draft should specify the following:

- Use of DF election to determine primary/backup (curious how it is done for plain old EVPN), including backup DF concept
- Only the current DF should set the P-bit
- Only the current BDF should set the B-bit
- P=B=1 is invalid
- There could be transient situations where multiple P=1 and/or multiple B=1 routes co-exist. Some heuristic should be used to pick which one to use to avoid traffic discard at the other end.

For the P=B=0 case - the receiving router can't use it anyway until its P/B is changed, right? If you want to use it as 2nd backup, the spec needs to point it out.

Thanks.

Jeffrey

> -----Original Message-----
> From: Rabadan, Jorge (Nokia - US) [mailto:[e-mail]]
> Sent: Thursday, May 12, 2016 2:28 PM
> To: Jeffrey (Zhaohui) Zhang <[e-mail]>; BESS <[e-mail]>
> Subject: Re: [bess] FW: WG Last Call (including implem status & shepherd)
> for draft-ietf-bess-evpn-vpws-03
> 
> HI Jeffrey,
> 
> 
> On 5/12/16, 8:13 PM, "BESS on behalf of EXT Jeffrey (Zhaohui) Zhang"
> <[e-mail] on behalf of [e-mail]> wrote:
> 
> >Hi Jorge,
> >
> >> >If a PE sends a route w/o setting P-bit, wouldn't that indicate it is
> a
> >> backup? Why would it bother sending the route if it does not want to be
> >> the backup?
> >>
> >> [JORGE] Let’s assume you have 3 PEs in ES-1 (PE1/2/3) that is a single-
> >> active ES. PE4 is a remote PE doing backup function.
> >
> >I assume PE4 is not "doing backup function" but is just one end of the PW.
> Or perhaps you mean it needs to decide which of PE1/2/3 is the
> primary/backup.
> 
> [JORGE2] by “doing backup function” I mean that PE4 sets up a destination
> to PE1 with a backup to PE2. Aliasing and Backup functions are effectively
> done by the remote PEs based on the information advertised by the PEs in
> the ES, right?
> 
> >
> >> - PE1 is the primary - the winner of the DF election. PE1 sends P=1
> >> - PE2 is the backup - the winner of the DF election (without PE1). PE2
> >> sends B=1.
> >
> >Hmm ... DF election is for BUM in RFC 7432 and never mentioned in this
> draft. If it is used for VPWS, it needs to be specified.
> 
> [JORGE2] In VPWS DF election is needed for single-active. The NDF will
> block the AC for the service. For all-active there is no DF, of course,
> since there is no BUM ;-)
> 
> >
> >The draft says multiple PEs could all set the P/B bits:
> >
> >   In multihoming single active scenario, a remote PE receiving P=1 from
> >   more than one PE will select only one primary PE when forwarding
> >   traffic. A remote PE receiving B=1 from more than one PE will select
> >   only one backup PE. A remote PE MUST receive P=1 from at least one PE
> >   before forwarding traffic.
> >
> >And the decision is based on the receiving PE, not on DF election?
> 
> [JORGE] This is BGP, there are always transient situations, so it may
> happen that even in single active you get more than one P=1 or B=1, so you
> need to say what to do in that case, right? In a steady situation, in
> single-active MH, only the DF MUST send the P=1 since it is only the DF
> the one unblocking the AC in the ES.
> 
> >
> >> - PE3 then sends P=B=0. This indicates that PE3 is neither primary nor
> >> backup, but the AC is active (if it was oper-down, it would withdraw
> the
> >> route).
> >
> >What purpose does the route from PE3 serve?
> 
> [JORGE] since the AC is oper-up on PE3, PE3 needs to send the route. At
> PE4 you know that PE3’s AC in the ES is good, but simply neither primary
> nor backup.
> 
> >
> >> - In case of a failure on PE1, PE2 will activate its AC on ES-1 since
> he
> >> wins the new DF election.
> >
> >What does "activate" involve exactly? Just re-advertise the route with P-
> bit set and B-bit cleared?
> 
> [JORGE] unblock Tx/Rx on the AC… and later readvertise the route with the
> new bits.
> 
> >
> >> At the same time, PE4 can immediately send
> >> traffic to PE2 as soon as PE1 withdraws the AD routes. PE4 does not
> even
> >> need to wait for the confirmation that PE2 is the new DF. This backup
> >> mechanism speeds up convergence big time.
> >> - The mechanism above does not work if PE2 and PE3 send both the same
> >> P=B=0, since it case of PE1 failure, PE4 could not send any traffic (it
> >> does not know who is the active one in the ES) and would need to wait
> for
> >> PE2 to send P=1.
> >
> >I don't see any use of the route from PE3. The draft seems to say that
> both PE1 and PE2 can set the P bit initially and PE4 will pick one to send
> traffic to. If it picks PE1 and then PE1 withdraws the route, PE4 will
> then pick PE2?
> 
> [JORGE] This is single active, so only one can transmit/receive to/from
> the CE. So only that one should send P=1.
> 
> >
> >Jeffrey
> >_______________________________________________
> >BESS mailing list
> >[e-mail]
> >https://www.ietf.org/mailman/listinfo/bess