Hi Stephane,
We'll modify the text based on your comments below.
Also for the benefit of the entire WG, based on Stephane's (and others)
feedback, we will add an indication of the support of the ac-df draft in the DF
Election extended community. The DF Election extended community is used by a
number of drafts, so it is important that the WG is aware of this. The proposed
encoding of the DF Election extended community is as follows:
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Type=0x06 | Sub-Type(0x06)| DF Type | Bitmap |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Reserved = 0 | DF Preference (2 octets) |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Where the following fields are defined as follows:
o DF Type can have the following values already used in some drafts:
- Type 0 - Default, mod based DF election as per RFC7432.
- Type 1 - HRW algorithm as per [DF Election draft]
- Type 2 - Preference algorithm [Preference draft]
- Type 3 - BW DF Election [BW DF election draft]
o Bitmap field defines 'capabilities' that may be supported with multiple
types. Some examples already used in some drafts:
·Bit 24 – DP (don’t preempt – Pref draft)
·Bit 25 – AC-influenced Election (ac-df draft)
·Bit 26 – handshake (handshake draft)
o The DF Type determines if the reserved fields are used or not. For example,
if DF type=2, the last 2 octets will encode the preference value used in the
preference draft.
The DF Election extended community will be defined in one of the drafts (TBD)
without defining the values above or the DF Preference. This draft will set up
the IANA registry for DF Types and Bitmap. Then each individual draft will
request IANA its own type or bit in the bitmap, and will specify if the new
type or capability can be supported with other types.
Please let us know if you have any feedback/questions.
Thank you.
Jorge
-----Original Message-----
From: "[email protected]" <[email protected]>
Date: Wednesday, January 31, 2018 at 9:39 AM
To: "Rabadan, Jorge (Nokia - US/Mountain View)" <[email protected]>,
"[email protected]"
<[email protected]>
Cc: "[email protected]" <[email protected]>
Subject: RE: Shepherd's review of draft-ietf-bess-evpn-ac-df
Hi Jorge,
Thanks for your answers.
Pls find more inline [SLI2]
-----Original Message-----
From: Rabadan, Jorge (Nokia - US/Mountain View)
[mailto:[email protected]]
Sent: Tuesday, January 30, 2018 16:20
To: LITKOWSKI Stephane OBS/OINIS;
[email protected]
Cc: [email protected]
Subject: Re: Shepherd's review of draft-ietf-bess-evpn-ac-df
Hi Stephane,
Thank you for your review.
Please see in-line and let me know what you think.
Thanks.
Jorge
-----Original Message-----
From: "[email protected]" <[email protected]>
Date: Tuesday, January 30, 2018 at 10:35 AM
To: "[email protected]"
<[email protected]>
Cc: "[email protected]" <[email protected]>
Subject: Shepherd's review of draft-ietf-bess-evpn-ac-df
Resent-From: <[email protected]>
Resent-To: <[email protected]>, <[email protected]>,
<[email protected]>, <[email protected]>, <[email protected]>,
<[email protected]>
Resent-Date: Tuesday, January 30, 2018 at 10:35 AM
Hi,
As shepherd of this document, please find below my comments.
IMO, this is a very useful proposal. The document is quite easy to
understand with a good illustrated example.
[JORGE] that's good, thanks.
Overall comments:
- I would encourage to have an acronym section containing all
abbreviations and the associated expansion. That could be a good best practice.
[JORGE] ok, done.
- I'm wondering why this document intended status is Informational. It
looks to fill a major (IMO) miss from the basic specification and I would be
happy to see it as a standard track document. Even if there is no protocol
extension involved, I think that standard track could make sense. Was this
already discussed ?
[JORGE] the reasons why the authors thought of Informational were because
a) it does not define any protocol extension and b) it is backwards compatible
with RFC7432, in the sense that a PE supporting this document and a PE
supporting RFC7432 could be put in the same Ethernet Segment and it would all
work (no loops or packet duplication). That may not necessarily be the case
with more than 2 PEs. Nevertheless the document does not recommend to do it in
any case. I am personally happy to change it to Standards Track if the rest of
the co-authors and the WG don't have any issues with it.
- I'm not sure that we can say that this procedure is backward
compatible. I agree that there is no protocol extension involved but as it is
mentioned, we cannot mix PEs running the new procedure and PE running the old
procedure. This must be ensured by the operator. Wouldn't it be better to add a
flag/attribute to announce that a PE is capable to run this procedure ? Thus,
when PEs run the svc carving algo, if they know that all the PEs are capable of
this procedure, and they can all run it automatically. If there is one or more
PE in the set that is not capable, they fallback to the regular procedure.
[JORGE] As discussed above, we wanted to make it backwards compatible in a
simple case. E.g. say PE1 and PE2 are in the same ES. PE1 supports this
document and PE2 does not. In this case, link/node failures are covered by the
baseline RFC7432. Logical AC failures don't change the RFC7432 overall
behavior, for instance:
a) a logical AC failure on PE1, will not change the DF, just like in
RFC7432.
b) a logical AC failure on PE2 makes PE1 take over, but, since PE2 can't
forward traffic anyway, there are no loops/duplication and everything works.
[SLI2] Being backward compatible on a simple case is IMO not enough to
declare the solution as backward compatible. As you mentioned if more than two
PEs are involved, we can fall into issues and human care are required. I don't
consider this as backward compatible.
- I would be in favor of having of deployment considerations section
that deals with the backward compatibility and how to deploy the solution.
[JORGE] We can add the section and the above example to clarify the
"backwards compatibility" concept.
[SLI2] That's good thanks. We just need to agree on the content :)
- There are too much authors in the Author list. Would it be possible
to reduce it ?
Problem statement:
"an ES route withdrawn will make...".
[SLI] I have a doubt here, would it be "and ES route withdrawal will
make" or "a withdrawn ES route" ?
[JORGE] fixed to "an ES route withdrawal". Thx.
Section 2.1
" After running the service-carving DF election algorithm"
[SLI] Could you mention that you refer to the existing algorithm from
RFC7432 ?
[JORGE] done. Thanks.
Section 2.2/2.3:
- Would it be possible to collapse the two cases in a single procedure
update ?
- I do not like to mix examples with a procedure update description. I
would rather be in favor of focusing on what is changing compared to RFC7432.
For instance, "The step 3 is changed as follows:". Keeping the example is
really good, so after describing the procedure, it makes sense to run it
through the example.
[JORGE] good point. I changed the text as follows:
"For the specific case of VLAN-aware bundle services, the DF election
will be influenced by the update/withdraw of "any" of the Ethernet A-
D per EVI routes in the <ESI,EVI>. Step 5 and 6 in section 3.2 are
therefore modified as follows:
5. When electing the DF for a given EVI, a PE will not be considered
candidate until "all" the Ethernet A-D per EVI routes for the
<ESI,EVI> have been received from that PE. In other words, all the
ACs on the <ESI,EVI> for a given PE must be UP so that the PE is
considered as candidate for a given EVI.
6. Once the PEs with missing Ethernet A-D per EVI routes for a given
EVI have been eliminated from the candidate list, the DF Election
can be applied for the remaining N candidates.
For example, assuming three bridge tables in PE-1 for the same MAC-
VRF (each one associated to a different Ethernet Tag), PE-1 will
advertise three Ethernet A-D per EVI routes for <ESI12,EVI1>. Each of
the three routes will indicate the status of each AC in <ESI12,EVI1>.
PE-1 will be considered as a valid candidate PE for DF election as
long as the three routes are active. If PE-1 withdraws one or more of
the Ethernet A-D per EVI routes for <ESI12,EVI1>, the PEs in ESI12
will not consider PE-1 as a suitable DF candidate for <ESI12,EVI1>."
[SLI2] Thanks for the proposal but it does not fully address my comment.
My point was to try to make section 2.2 and 2.3 as an update of the
section 8.5 of RFC7432 (which step in RFC7432 procedure is modified).
The text appears as redefining a full procedure while in reality only one
step is modified.
Brgds,
Stephane Litkowski
_________________________________________________________________________________________________________________________
Ce message et ses pieces jointes peuvent contenir des informations
confidentielles ou privilegiees et ne doivent donc
pas etre diffuses, exploites ou copies sans autorisation. Si vous avez
recu ce message par erreur, veuillez le signaler
a l'expediteur et le detruire ainsi que les pieces jointes. Les
messages electroniques etant susceptibles d'alteration,
Orange decline toute responsabilite si ce message a ete altere, deforme
ou falsifie. Merci.
This message and its attachments may contain confidential or privileged
information that may be protected by law;
they should not be distributed, used or copied without authorisation.
If you have received this email in error, please notify the sender and
delete this message and its attachments.
As emails may be altered, Orange is not liable for messages that have
been modified, changed or falsified.
Thank you.
_________________________________________________________________________________________________________________________
Ce message et ses pieces jointes peuvent contenir des informations
confidentielles ou privilegiees et ne doivent donc
pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu
ce message par erreur, veuillez le signaler
a l'expediteur et le detruire ainsi que les pieces jointes. Les messages
electroniques etant susceptibles d'alteration,
Orange decline toute responsabilite si ce message a ete altere, deforme ou
falsifie. Merci.
This message and its attachments may contain confidential or privileged
information that may be protected by law;
they should not be distributed, used or copied without authorisation.
If you have received this email in error, please notify the sender and
delete this message and its attachments.
As emails may be altered, Orange is not liable for messages that have been
modified, changed or falsified.
Thank you.
_______________________________________________
BESS mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/bess
