Enke Chen | 2 Mar 2008 06:30
Picon
Favicon

comments on draft-chakrabarti-idr-rfc4893-mod-00.txt

Hi, folks:

I am posting my comments to the mailing list as I will not be at the 
upcoming
IETF meeting.

-) On Issue - 1

   IMO RFC4893 is pretty clear in stating that the Capability "carries 
the 4-octet
   Autonomous System number of the speaker in the Capability Value 
field". What
   else needs to be said?

-) On Issue - 2:

   The issues with having islands of the same AS number are well understood.
   Please see "RFC 2270: Using a Dedicated AS for Sites Homed to a 
Single Provider".

   We could add the reference when there is ever sufficient reason to 
update the document
   in the future.

-) On Issue - 3:

    The mechanism described in RFC 4893 is designed to avoid the need for a
    second transition for removing one of the attributes from the global 
routing
    system.
(Continue reading)

Samita Chakrabarti | 5 Mar 2008 03:21
Favicon

Re: comments on draft-chakrabarti-idr-rfc4893-mod-00.txt

Hi Enke,

 

Thanks for your input. Please see my responses below.

 

>-----Original Message-----

>From: idr-bounces <at> ietf.org [mailto:idr-bounces <at> ietf.org] On Behalf Of Enke

>Chen

>Sent: Saturday, March 01, 2008 9:31 PM

>To: idr

>Subject: [Idr] comments on draft-chakrabarti-idr-rfc4893-mod-00.txt

>

 

>Hi, folks:

>

 

>I am posting my comments to the mailing list as I will not be at the

>upcoming

>IETF meeting.

>

 

>;-) On Issue - 1

>

 

>   IMO RFC4893 is pretty clear in stating that the Capability "carries

>the 4-octet

>   Autonomous System number of the speaker in the Capability Value

>field". What

>   else needs to be said?

[SC>]  The issue-1 describes the need for clarification for implementation on OPEN message processing when AS number is present in both capability message and in “MY AS Number” fields.

IMO, The document should clarify that AS Number present in Capability message is used for the AS number and “My AS number” field is ignored. From RFC 4893, it is assumed that nodes with 4byte-mappable 2-byte AS number can advertise capability (AS4_PATH) as well as nodes with unmappable 4-byte AS numbers. In case of un-mappable 4-byte AS numbers,the “My AS number” field is AS_TRANS.

 

>

 

>;-) On Issue - 2:

>

 

>   The issues with having islands of the same AS number are well understood.

>   Please see "RFC 2270: Using a Dedicated AS for Sites Homed to a

>Single Provider".

>

 

>   We could add the reference when there is ever sufficient reason to

>update the document

>   in the future.

[SC>]  Yes. It should be updated. IMHO, it might be a good idea to clarify and update the document now before the deployment starts and face interoperability issues. Are there many implementations in AS4_PATH that have been interop tested ?

>

 

>;-) On Issue - 3:

>

 

>    The mechanism described in RFC 4893 is designed to avoid the need for a

>    second transition for removing one of the attributes from the global

>routing

>    system.

>

 

>    The proposal in Sect. 6.1 is backwards, and would require exactly

>the second

>    transition that we have worked very hard to avoid. The proposal is

>also not

>    realistic as there is simply no easy way to know when the transition is

>    "complete".

[SC>] I don’t think RFC 4893 can either tell if the transition is complete or not by looking at the AS_PATH, since AS_PATH will always contain 4-byte ASes between two NEW BGP speakers. At the border NEW BGP speaker, it gets the AS_PATH with 2-byte ASes from the OLD BGP speaker, but there is no way to know how many ASes are new by looking at the AS_PATH because some speakers on the path may be 4-byte capable but they have a 2-byte mappable AS number.

>

 

[SC>] Here is a brief analysis:

 

RFC 4893 style:

 Pro: A new BGP speaker can dynamically convert 2-byte AS number to 4-byte AS-number in AS_PATH at the boundary.

 

Con: This requires extra processing of AS_PATH, AS_AGGREGATE conversion when there is a OLD BGP peer on one side and a NEW BGP peer on the other side. 4-byte AS number migration will take time, thus during the long period of transition, the border New BGP speaker will have to do this extra processing. A new BGP speaker sometimes will process 4byte AS number in AS_PATH and sometimes 2-bytes depending on whether it is coming from a NEW speaker or OLD speaker. This overloading of AS_PATH usage can be error-prone.

 

Proposed style in draft-chakrabarti-idr-rfc4893-mod-00.txt:

Pro: A NEW BGP speaker with 4-byte AS number always includes AS4_PATH

   attribute containing the extended 4-byte AS number.  If the AS number

   is 2-byte mappable, then it adds the corresponding 2-byte mapped AS

   number in the AS_PATH attribute, otherwise it uses AS_TRANS as the AS

   number in the corresponding AS_PATH attribute.  Thus the NEW BGP

   speaker will always have AS4_PATH and a corresponding AS_PATH

   attribute.  No conversion is needed at the transition boundary. Both AS_PATH and AS4_PATH will co-exist during the transition period.

 

   When transition is over, a BGP speaker can use AS4_PATH only. It is a simple concept and no overloading.

 

 

Con: Looking at AS_PATH or AS4_PATH info, one can not tell if the transition has already happened in the Internet. But how important is it to know this way? Was this a design goal ?

 

>;-) On Issue - 4:

>

 

>    The reservation of AS_TRANS is relatively new. There are tons of old

>speakers in

>    the field that would happily peer using the value if so configured.

>We can not

>    mandate what an old speaker should or should not do as they are

>already there

>    in the field :-)

>

 

>    Also note that AS_TRANS is not the only reserved AS number.

>

 

>    It is not clear if there is any practical value for introducing the

>notification.

>

 

[SC>] I talked with a few people at NANOG who experienced that AS_TRANS value was used often as AS number in the current BGP deployment. We can not expect that every operator or administrator is fully aware of reserved AS_TRANS value for this special purpose. Thus the notification message from NEW BGP speaker to OLD BGP speaker who uses AS_TRANS value as “My AS number” field in the OPEN message is very helpful during transition.

 

[SC>] besides, the transition section of RFC4893 should also mention the order of AS number transition of PE/CE routers.Some folks think that PE routers should be transitioned first (if not at the same time) in order to take care of AS override function at the egress path. [ it is not part of my draft yet ]

 

Regards,

-Samita

 

 

_______________________________________________
Idr mailing list
Idr <at> ietf.org
https://www.ietf.org/mailman/listinfo/idr
Enke Chen | 5 Mar 2008 03:58
Picon
Favicon

Re: comments on draft-chakrabarti-idr-rfc4893-mod-00.txt

Hi, Samita:

Please see my comments inlined.

Samita Chakrabarti wrote:
>
> Hi Enke,
>
> Thanks for your input. Please see my responses below.
>
> >-----Original Message-----
>
> >From: idr-bounces <at> ietf.org [mailto:idr-bounces <at> ietf.org] On Behalf Of 
> Enke
>
> >Chen
>
> >Sent: Saturday, March 01, 2008 9:31 PM
>
> >To: idr
>
> >Subject: [Idr] comments on draft-chakrabarti-idr-rfc4893-mod-00.txt
>
> >
>
> >Hi, folks:
>
> >
>
> >I am posting my comments to the mailing list as I will not be at the
>
> >upcoming
>
> >IETF meeting.
>
> >
>
> >-) On Issue - 1
>
> >
>
> > IMO RFC4893 is pretty clear in stating that the Capability "carries
>
> >the 4-octet
>
> > Autonomous System number of the speaker in the Capability Value
>
> >field". What
>
> > else needs to be said?
>
> */ [SC>] The issue-1 describes the need for clarification for 
> implementation on OPEN message processing when AS number is present in 
> both capability message and in “MY AS Number” fields./*
>

The text in the document is very specific and clear.

> */ /*
>
> */ IMO, The document should clarify that AS Number present in 
> Capability message is used for the AS number and “My AS number” field 
> is ignored. From RFC 4893, it is assumed that nodes with 
> 4byte-mappable 2-byte AS number can advertise capability (AS4_PATH) as 
> well as nodes with unmappable 4-byte AS numbers. In case of 
> un-mappable 4-byte AS numbers,the “My AS number” field is AS_TRANS./*
>

Again, the document is clear about using AS_TRANS in the "My AS number" 
field:

   To represent 4-octet AS numbers (which are not mapped from 2-octets)
   as 2-octet AS numbers in the AS path information encoded with 2-octet
   AS numbers, this document reserves a 2-octet AS number.  We denote
   this special AS number as AS_TRANS for ease of description in the
   rest of this specification.  This AS number is also placed in the "My
   Autonomous System" field of the OPEN message originated by a NEW BGP
   speaker, if the speaker does not have a (globally unique) 2-octet AS
   number.

> */ /*
>
> */ /*
>
> >
>
> >-) On Issue - 2:
>
> >
>
> > The issues with having islands of the same AS number are well 
> understood.
>
> > Please see "RFC 2270: Using a Dedicated AS for Sites Homed to a
>
> >Single Provider".
>
> >
>
> > We could add the reference when there is ever sufficient reason to
>
> >update the document
>
> > in the future.
>
> */ [SC>] Yes. It should be updated. IMHO, it might be a good idea to 
> clarify and update the document now before the deployment starts and 
> face interoperability issues. Are there many implementations in 
> AS4_PATH that have been interop tested ?/*
>

Just FYI - there have been multiple implementations and deployments 
since 2004.

In terms of transition, please see Section 6. It assumes a reasonable 
way of transition:

   To simplify transition, this document assumes that an Autonomous
   System could start using a 4-octet AS number only after all the BGP
   speakers within that Autonomous System have been upgraded to support
   4-octet AS numbers.

> */ /*
>
> >
>
> >-) On Issue - 3:
>
> >
>
> > The mechanism described in RFC 4893 is designed to avoid the need for a
>
> > second transition for removing one of the attributes from the global
>
> >routing
>
> > system.
>
> >
>
> > The proposal in Sect. 6.1 is backwards, and would require exactly
>
> >the second
>
> > transition that we have worked very hard to avoid. The proposal is
>
> >also not
>
> > realistic as there is simply no easy way to know when the transition is
>
> > "complete".
>
> */ [SC>] I don’t think RFC 4893 can either tell if the transition is 
> complete or not by looking at the AS_PATH, since AS_PATH will always 
> contain 4-byte ASes between two NEW BGP speakers. At the border NEW 
> BGP speaker, it gets the AS_PATH with 2-byte ASes from the OLD BGP 
> speaker, but there is no way to know how many ASes are new by looking 
> at the AS_PATH because some speakers on the path may be 4-byte capable 
> but they have a 2-byte mappable AS number. /*
>
> >
>
> */ [SC>] Here is a brief analysis: /*
>
> */ /*
>
> */ RFC 4893 style: /*
>
> */ Pro: A new BGP speaker can dynamically convert 2-byte AS number to 
> 4-byte AS-number in AS_PATH at the boundary. /*
>
> */ /*
>
> */ Con: This requires extra processing of AS_PATH, AS_AGGREGATE 
> conversion when there is a OLD BGP peer on one side and a NEW BGP peer 
> on the other side. 4-byte AS number migration will take time, thus 
> during the long period of transition, the border New BGP speaker will 
> have to do this extra processing. A new BGP speaker sometimes will 
> process 4byte AS number in AS_PATH and sometimes 2-bytes depending on 
> whether it is coming from a NEW speaker or OLD speaker. This 
> overloading of AS_PATH usage can be error-prone. /*
>
> */ /*
>
> */ Proposed style in draft-chakrabarti-idr-rfc4893-mod-00.txt: /*
>
> */ Pro: /* A NEW BGP speaker with 4-byte AS number always includes 
> AS4_PATH
>
> attribute containing the extended 4-byte AS number. If the AS number
>
> is 2-byte mappable, then it adds the corresponding 2-byte mapped AS
>
> number in the AS_PATH attribute, otherwise it uses AS_TRANS as the AS
>
> number in the corresponding AS_PATH attribute. Thus the NEW BGP
>
> speaker will always have AS4_PATH and a corresponding AS_PATH
>
> attribute. No conversion is needed at the transition boundary. Both 
> AS_PATH and AS4_PATH will co-exist during the transition period.
>
> When transition is over, a BGP speaker can use AS4_PATH only. It is a 
> simple concept and no overloading.
>
> */ /*
>
> */ Con: Looking at AS_PATH or AS4_PATH info, one can not tell if the 
> transition has already happened in the Internet. But how important is 
> it to know this way? Was this a design goal ? /*
>
>

This is a moot point at this time. After years of implementation and 
deployment, there can not be any change to the update generation. 
Besides what you proposed is something that was discussed and not 
pursued due to the major flaw I pointed out.

> >-) On Issue - 4:
>
> >
>
> > The reservation of AS_TRANS is relatively new. There are tons of old
>
> >speakers in
>
> > the field that would happily peer using the value if so configured.
>
> >We can not
>
> > mandate what an old speaker should or should not do as they are
>
> >already there
>
> > in the field :-)
>
> >
>
> > Also note that AS_TRANS is not the only reserved AS number.
>
> >
>
> > It is not clear if there is any practical value for introducing the
>
> >notification.
>
> >
>
> */ [SC>] I talked with a few people at NANOG who experienced that 
> AS_TRANS value was used often as AS number in the current BGP 
> deployment. We can not expect that every operator or administrator is 
> fully aware of reserved AS_TRANS value for this special purpose. Thus 
> the notification message from NEW BGP speaker to OLD BGP speaker who 
> uses AS_TRANS value as “My AS number” field in the OPEN message is 
> very helpful during transition./*
>

Sorry to say that you might be misinformed of the use of AS 23456. In 
the public internet, one must use an AS number allocated by IANA. 
Otherwise there are consequences, including partial connectivities ....

The allocation and management of AS numbers are outside the scope of the 
document.

> */ /*
>
> */ /*
>
> */ [SC>] besides, the transition section of RFC4893 should also 
> mention the order of AS number transition of PE/CE routers.Some folks 
> think that PE routers should be transitioned first (if not at the same 
> time) in order to take care of AS override function at the egress 
> path. [ it is not part of my draft yet ]/*
>

Please see Section 6 (Transition) of the document. The document assumes 
(probably should recommends) that the provider networks are upgraded 
first before connecting non-mappable 4byte AS peers.

Regards, -- Enke

> */ /*
>
> */
> /*
>
>
_______________________________________________
Idr mailing list
Idr <at> ietf.org
https://www.ietf.org/mailman/listinfo/idr

Geoff Huston | 5 Mar 2008 06:09
Favicon

Re: comments on draft-chakrabarti-idr-rfc4893-mod-00.txt

>>-) On Issue - 3:
>> 
>>    The mechanism described in RFC 4893 is designed to avoid the need for a
>>    second transition for removing one of the attributes from the global routing
>>    system.
>> 
>>    The proposal in Sect. 6.1 is backwards, and would require exactly the second
>>    transition that we have worked very hard to avoid. The proposal is also not
>>    realistic as there is simply no easy way to know when the transition is
>>    "complete".

...

> 
>    When transition is over, a BGP speaker can use AS4_PATH only. It is a 
> simple concept and no overloading.

But there is no way for any individual BGP speak to know when transition 
is "over". I have to agree with Enke in observing that the proposal is 
indeed backwards.

>>-) On Issue - 4:
> 
>> 

> */[SC>] I talked with a few people at NANOG who experienced that 
> AS_TRANS value was used often as AS number in the current BGP 
> deployment. We can not expect that every operator or administrator is 
> fully aware of reserved AS_TRANS value for this special purpose.

I'm sorry, but this is a specious line of reasoning.

Overall I see noting of use in this draft to warrant modification of RFC4893

   Geoff

_______________________________________________
Idr mailing list
Idr <at> ietf.org
https://www.ietf.org/mailman/listinfo/idr

Brian Dickson | 5 Mar 2008 06:51

draft-dickson-idr-second-best-backup-02

Brian Dickson wrote:
> draft-dickson-idr-second-best-backup-01 has been posted to the expected place:
> http://www.ietf.org/internet-drafts/draft-dickson-idr-second-best-backup-01.txt
>
>   

Version -02 will be posted when postings are permitted again, after Philly.

> There have been some rather substantial changes made to parts of it, especially as far as how withdrawals
are structured, and matched to updates.
>
> Also, the capabilities have been rolled together, since one type (backup) relies on the other to achieve
the goal of preventing wedgies.
>
> For IETF-71, I'll be doing some comparisons between this and other proposals (e.g. the walton add-paths one).
>
> In the meantime, feedback is welcome, although no updates will occur between now and the impending IETF
meeting. :-)
>   

I exaggerated a bit, unfortunately.

After some discussions (kudos to Joel Halpern), I've managed to strip a 
lot of unnecessary elements from the proposal, and cleaned up the text.

I think it will be much clearer, and be easier to read or even skim 
over, now.

Because of that, and to be kind to those like myself who read IDs only 
shortly before or during the IETF meetings, I'm posting this to the list.

Please accept my apologies for not doing so sooner, but better a little 
late and much cleaner, is a good way to go IMHO.

(The "backup" stuff has been removed, and will instead be proposed as a 
new "well-known" community value instead.)

Brian Dickson

idr                                                           B. Dickson
Internet-Draft                                       Afilias Canada, Inc
Expires: August 4, 2008                                    February 2008

       Enhanced BGP Capabilities for Exchanging Second-Best Paths
                draft-dickson-idr-second-best-backup-02

Status of this Memo

   By submitting this Internet-Draft, each author represents that any
   applicable patent or other IPR claims of which he or she is aware
   have been or will be disclosed, and any of which he or she becomes
   aware will be disclosed, in accordance with Section 6 of BCP 79.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF), its areas, and its working groups.  Note that
   other groups may also distribute working documents as Internet-
   Drafts.

   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any
   time.  It is inappropriate to use Internet-Drafts as reference
   material or to cite them other than as "work in progress."

   The list of current Internet-Drafts can be accessed at
   http://www.ietf.org/ietf/1id-abstracts.txt.

   The list of Internet-Draft Shadow Directories can be accessed at
   http://www.ietf.org/shadow.html.

   This Internet-Draft will expire on August 4, 2008.

Copyright Notice

   Copyright (C) The IETF Trust (2008).

Dickson                  Expires August 4, 2008                 [Page 1]

Internet-Draft         BGP Second-Best and Back-up         February 2008

Abstract

   This Internet Draft describes an enhanced way to exchange prefix
   information, to permit two copies of a prefix with different paths to
   be announced and withdrawn.

   This negotiated capability facilitates faster local (inter-AS) and
   global (intra-AS) convergence, reduces path-hunting, improves route-
   reflector behaviour, including eliminating persistent oscillations.

   Additional prefix instances have a new optional BGP attribute to
   control path selection.

   Withdrawal of prefixes will require a slight modification to
   disambiguate prefix instances.

   Benefits are seen both when deployed intra-AS, and on inter-AS
   peering.

Dickson                  Expires August 4, 2008                 [Page 2]

Internet-Draft         BGP Second-Best and Back-up         February 2008

Author's Note

   This Internet Draft is intended to result in this draft or a related
   draft(s) being placed on the Standards Track for idr.

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
   document are to be interpreted as described in [4].

   Intended Status: Proposed Standard.

Table of Contents

   1.  Background . . . . . . . . . . . . . . . . . . . . . . . . . .  4
     1.1.  The Best Path Chaining and the Best Path Tree  . . . . . .  4
     1.2.  The Withdrawal Problem . . . . . . . . . . . . . . . . . .  4
     1.3.  The Uniqueness Property  . . . . . . . . . . . . . . . . .  5
   2.  Proposed Changes . . . . . . . . . . . . . . . . . . . . . . .  6
     2.1.  New Negotiated Option: USE_SECOND_BEST . . . . . . . . . .  6
     2.2.  New Optional Path Attribute: SECOND_BEST . . . . . . . . .  6
     2.3.  New Update Format  . . . . . . . . . . . . . . . . . . . .  7
     2.4.  New Withdraw Format  . . . . . . . . . . . . . . . . . . .  7
   3.  Modifications to BGP Behavior  . . . . . . . . . . . . . . . .  9
     3.1.  Changes to Path Selection Rules  . . . . . . . . . . . . .  9
     3.2.  Second Best - Basic Method . . . . . . . . . . . . . . . . 10
     3.3.  Second Best - Route Reflector  . . . . . . . . . . . . . . 10
     3.4.  Second Best - Inter-AS Hybrid Method . . . . . . . . . . . 10
     3.5.  IBGP vs EBGP . . . . . . . . . . . . . . . . . . . . . . . 10
   4.  Implementation Guidelines  . . . . . . . . . . . . . . . . . . 12
   5.  Security Considerations  . . . . . . . . . . . . . . . . . . . 14
   6.  IANA Considerations  . . . . . . . . . . . . . . . . . . . . . 15
   7.  Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 16
   8.  References . . . . . . . . . . . . . . . . . . . . . . . . . . 17
     8.1.  Normative References . . . . . . . . . . . . . . . . . . . 17
     8.2.  Informative References . . . . . . . . . . . . . . . . . . 17
   Appendix A.  Path-Hunting Examples . . . . . . . . . . . . . . . . 18
   Appendix B.  Persistent Oscillation Examples . . . . . . . . . . . 19
   Author's Address . . . . . . . . . . . . . . . . . . . . . . . . . 21
   Intellectual Property and Copyright Statements . . . . . . . . . . 22

Dickson                  Expires August 4, 2008                 [Page 3]

Internet-Draft         BGP Second-Best and Back-up         February 2008

1.  Background

   Even when all the best current practises are observed, operational
   problems may be experienced when running a BGP network.

   These include slow convergence due to "path-hunting" and persistant
   oscillations [1].

   Standardization of MRAI timers helps path-hunting, and oscillations
   can be worked around with RFC 5004 [3].

   However, both of these RFCs identify the above issues as needing
   further work.

1.1.  The Best Path Chaining and the Best Path Tree

   In a stable system of BGP speakers, for every given prefix, the
   selected best paths should form a spanning tree.  At each node, the
   best path selected points further up the tree.  The root of the tree
   is the destination, i.e. the originator of the prefix.  The path from
   any leaf to the root forms a "chain" of best paths.

   There are any number of ways that path attributes may be modified
   over time, at arbitrary places in this tree.  When this happens,
   individual segments of the tree may conceptually "stretch" or
   "shrink".  These changes may have no effect on the overall set of
   choices of best path, or they may cause a cascade effect "below" that
   point in the tree, with nodes migrating to new locations in a new
   version of the tree.

   However, each node makes its choice of best path locally, and every
   time a node changes its selection of best path, that change is
   visible to its peers, and may in turn affect their own choice of best
   path.  This propogation of changes is not instantaneous, and owing to
   the non-tree-like nature of the actual connectivity between nodes,
   can and does result in race conditions.

   Depending on connectivity, peering policy, and initial conditions,
   the behavior may border on that of systems best describe through
   chaos theory.  The time to reach a stable state, while generally
   bounded, is often far from fast, not necessarily predictable, and not
   necessarily consistent.

1.2.  The Withdrawal Problem

   Under normal circumstances, a change in attributes for a prefix will
   "flow" along the tree of best paths, without disrupting the structure
   of the tree itself signficantly.  Even when a node selects a new best

Dickson                  Expires August 4, 2008                 [Page 4]

Internet-Draft         BGP Second-Best and Back-up         February 2008

   path (and thus re-attaches itself to the tree in a new location), it
   typically will continue to pass the new attributes along the branch
   of the tree for which it is the root.

   However, under certain circumstances, its choice of new best path,
   requires it to WITHDRAW the prefix from those peers, and effectively
   sever the branch.  It is in the after-effects of this truncation that
   much of the path-hunting behavior gets triggered.

   When a withdrawal effectively severs a branch of the tree, all the
   nodes on the tree will need to find new paths to the root.  The
   problem is, that it takes some time for them to learn this fact.

   In the mean time, the nodes in the severed branch may continue to
   use, and propogate, paths that are technically infeasible.

   The idea is to fast-track the flooding of the infeasibility of paths
   throughout all parts of the tree below a given link, so as to
   minimize the use of infeasible paths.

1.3.  The Uniqueness Property

   Currently, for each prefix, only one path for that prefix is ever
   announced from one peer to another (ignoring Route Reflectors).
   Because of this property, uniqueness, a withdrawal on a prefix does
   not require path information.  This also means that a change of best
   path is accomplished via an update for a prefix with the new path
   information.

   If, however, more than one path for a given prefix were sent, then
   any attempt to withdraw a prefix+path would require some mechanism to
   distinguish between prefix instances.

   In an environment where multiple path announcments per prefix are
   possible, but only one "best" path per prefix is maintained, then two
   steps would be involved in changing the "best" path.  In no
   particular order, that would be the withdrawal of the old prefix+
   path, and the announcement of the new prefix+path.

Dickson                  Expires August 4, 2008                 [Page 5]

Internet-Draft         BGP Second-Best and Back-up         February 2008

2.  Proposed Changes

   What is being proposed is, maintaining a "top two" for each prefix,
   and sending both of these rather than just the "best" path.

   When either of the "top two" becomes infeasible, a withdrawal is
   sent.  If a withdrawal is received, it receives special fast-track
   handling, taking advantage of the "top two" information.  If either
   of the top two is affected by the withdrawal, it is immediately
   flooded to peers without doing a full BGP table walk.

   The supposition is that pruning all infeasible branches, while
   maintaining information on second-best paths, allows for fast removal
   of all best paths which are dependent on infeasible paths, and fast
   reconvergence with pre-computed alternate paths.  It is expected that
   the second-best mechanism should act as a stop-gap until, but not
   actually replace, full BGP table walking to generate a new set of
   "top two" paths.

2.1.  New Negotiated Option: USE_SECOND_BEST

   This is a new BGP Capabilities value, which can be optionally
   included in the capabilities negotiation.  The specific value is a
   code-point to be assigned by IANA.

   When negotiated:

   o  Update messages MUST be in the new format

   o  Updates without any of the new optional attributes are considered
      BEST

   o  For each prefix, at most one of each type (BEST and SECOND_BEST)
      may be sent

2.2.  New Optional Path Attribute: SECOND_BEST

   This is a new BGP Path Attribute type.  It MAY be used only if the
   USE_SECOND_BEST capability has been negotiated.  The type value is a
   new code point to be assigned by IANA.

   This is an Optional, Non-Transitive, Non-Extended, Non-Partial
   attribute.  All the "attr flag bits" (from BGP [2]) are zero.  The
   length is 1, and the value is 1.

Dickson                  Expires August 4, 2008                 [Page 6]

Internet-Draft         BGP Second-Best and Back-up         February 2008

2.3.  New Update Format

   Update messages are identical to existing format, with the exception
   of the new Withdrawal format, and the new optional Path Attribute
   (SECOND_BEST).  If BGP capability USE_SECOND_BEST has been
   negotiated, any Update MAY have a Path Attribute of SECOND_BEST.
   More than one instance of a given prefix, with distinct values of
   Path Attributes, MAY be sent between BGP speakers.

   At most two instances may be sent, specifically one of each
   combination of with/without SECOND_BEST.

   Two prefix paths are considered identical if they differ only in the
   presence or absence of the new attribute.  An Update which contains a
   path which differs from the previous path of that type (BEST or
   SECOND_BEST), will result in the path information for the prefix and
   type being modified.

2.4.  New Withdraw Format

   Since it is no longer possible to identify which instance of an
   prefix is affected by an update containing a withdrawal, a new format
   for Withdrawals is needed.  For simplicity of implementations, this
   consists of two Withdrawal sections, one for each of the types (BEST,
   SECOND_BEST).  They occur in REVERSE order, to simplify state
   transitions if/when a "BEST" path is withdrawn.  Each Withdrawal sub-
   section has the same format as the original Withdrawal section.

         +-----------------------------------------------------+
         |   Withdrawn Routes Length (2 octets)                |
         +-----------------------------------------------------+
         |   Withdrawn Routes (variable)                       |
         +-----------------------------------------------------+
         |   Total Path Attribute Length (2 octets)            |
         +-----------------------------------------------------+
         |   Path Attributes (variable)                        |
         +-----------------------------------------------------+
         |   Network Layer Reachability Information (variable) |
         +-----------------------------------------------------+

                                 Figure 1

   Withdrawn Routes Length:  This 2-octets unsigned integer indicates
      the total length of the Withdrawn Routes field in octets.  Its
      value allows the length of the Network Layer Reachability
      Information field to be determined, as specified below.

Dickson                  Expires August 4, 2008                 [Page 7]

Internet-Draft         BGP Second-Best and Back-up         February 2008

      A value of 0 indicates that no routes are being withdrawn from
      service, and that the WITHDRAWN ROUTES field is not present in
      this UPDATE message.

   Withdrawn Routes Field:  This field now consists of two sub-fields
      and their respective lengths.  The value for Withdrawn Routes
      Length above, must be the sum of the two lengths, plus 4 (the sum
      of the lengths of the Subfield Lengths).

      The format and sequence of the subfields is as follows:

      +----------------------------------------------------------------+
      |   Withdrawn SECOND_BEST Routes Length (2 octets)               |
      +----------------------------------------------------------------+
      |   Withdrawn SECOND_BEST Routes (variable)                      |
      +----------------------------------------------------------------+
      |   Withdrawn BEST Routes Length (2 octets)                      |
      +----------------------------------------------------------------+
      |   Withdrawn BEST Routes (variable)                             |
      +----------------------------------------------------------------+

                                   Figure 2

   Withdrawn Routes Subfield Lengths  These 2-octets unsigned integers
      indicates the total length of their respective Withdrawn Routes
      subfields in octets.

   Withdrawn Routes Subfields:  Each of these is a variable-length field
      that contains a list of IP address prefixes for the routes that
      are being withdrawn from service.  Each IP address prefix is
      encoded as a 2-tuple of the form <length, prefix>, whose fields
      are described below:

                     +---------------------------+
                     |   Length (1 octet)        |
                     +---------------------------+
                     |   Prefix (variable)       |
                     +---------------------------+

Dickson                  Expires August 4, 2008                 [Page 8]

Internet-Draft         BGP Second-Best and Back-up         February 2008

3.  Modifications to BGP Behavior

3.1.  Changes to Path Selection Rules

   The path selection rules for BGP (section 9.1.2.2 of BGP4 [2]) are
   changed as follows:

   o  The following rule is a modification to step (c).

      It MAY only be needed when the node is acting as a Route
      Reflector.  If a node is NOT a Route Reflector, a simplified
      modification (remove any paths marked SECOND_BEST) MAY be used.
      (This modification exists to resolve the Persistent Oscillation
      problem only.)

      The modification to step (c) is:

      Step (c) is first performed INCLUDING paths with SECOND_BEST.

      If, at the end of the first attempt at step (c), only paths with
      SECOND_BEST remain, re-run step (c), this time EXCLUDING the paths
      with SECOND_BEST.

      After this modified version of step (c), the remaining paths MUST
      NOT have the SECOND_BEST attribute.

      In other words, Step (c) MUST remove any SECOND_BEST paths.

   o  The remainder of the usual BGP path selection rules are applied as
      normal

   The path selection rules for "Second Best" path are as follows:

   o  The already-selected "best" path is removed from the set of paths
      to compare

   o  The SECOND_BEST entry for the prefix, from the same in-RIB-SB that
      the "best" path came from, is temporarily promoted to "best".

   o  The same rules are applied as for the "best" path

   o  The selected path is advertised with the attribute SECOND_BEST
      applied

   The prefix instances for consideration of second-best path are the
   REMAINDER of non-SECOND_BEST instances, and the SECOND_BEST instance
   received on the in-RIB from which the best path was selected (if one
   exists).  Only one SECOND_BEST instance received may be considered

Dickson                  Expires August 4, 2008                 [Page 9]

Internet-Draft         BGP Second-Best and Back-up         February 2008

   for the local (and out-RIB) SECOND_BEST path.

3.2.  Second Best - Basic Method

   Once the capabality for doing so has been negotiated between a pair
   of BGP speakers, each sends the best two paths for each prefix.  The
   path information will include the additional SECOND_BEST attribute on
   the second best path.

   When the current "best" path is withdrawn, the withdrawal MAY be
   propogated without having to perform a full BGP table path selection.
   The current "second best" path in the local-RIB is promoted to
   "best".  This is because the alternate candidates have already been
   evaluated and "second-best" has already been selected.

   Whenever an AS consists of a mesh of BGP speakers who have negotiated
   this capability, the withdrawal will propogate through the entire AS.
   This will either have no effect, or with a change in "best" without
   requiring non-local information to choose the new "best" path.

   The second-best path from a neighbor MUST ONLY be considered as a
   candidate for best path, when the previous best path from that
   neighbor is withdrawn.  When this occurs, the path in question is
   promoted to "best" status.

3.3.  Second Best - Route Reflector

   The "best" and "second best" are reflected.  The same mechanism is
   used for determining both best and second-best per prefix.  Updates
   must be reflected whenever the choice of either or both of the "best"
   or "second best" change.  Withdrawals may be propogated immediately.

3.4.  Second Best - Inter-AS Hybrid Method

   When a withdrawal of the current best path is received from a peer
   doing USE_SECOND_BEST, and the rules for sending updates require that
   an update for this prefix be sent to a peer who does not support
   USE_SECOND_BEST, the current second-best instance of the prefix is
   sent to that peer in an Update.  The neighbor does not need the
   withdrawal, since the new path replaces the old path.

3.5.  IBGP vs EBGP

   The same rules apply for EBGP->EBGP, EBGP->IBGP, IBGP->EBGP, and
   IBGP->IBGP.  If a particular peering has had USE_SECOND_BEST
   negotiated, then any update for a particular prefix that results in
   new selection of either or both of best and second-best, the new
   selections (and possible withdrawal of old selections) is sent to the

Dickson                  Expires August 4, 2008                [Page 10]

Internet-Draft         BGP Second-Best and Back-up         February 2008

   appropriate peers.

Dickson                  Expires August 4, 2008                [Page 11]

Internet-Draft         BGP Second-Best and Back-up         February 2008

4.  Implementation Guidelines

   In order to encourage effective implementation schemes, and to
   demonstrate some of the benefits of deployment, here are some
   suggestions for facilitating fast propogation of path changes, which
   are anticipated as improving behavior.  This applies in particular to
   Path Hunting issues.

   In-RIB-SB (many) -> RIB-SB -> out-RIB-SB
                       |   \
                       v    `-> out-RIB (to non-SB peers)
                       RIB -> FIB

   +----------+-------+--------+-----------------|
   |   PREFIX | IN-SB | OUT-SB | *PATH-info-ptr  |
   +----------+-------+--------+-----------------|

                                 Figure 4

   Where IN-SB and OUT-SB are 2-bit fields indicating what the
   SECOND_BEST (SB) state is (BEST, SECOND_BEST).  Valid values for
   IN-SB are "10" and "01"; for OUT-SB, are "00", "10", and "01".  IN-SB
   are the attributes received from a peer.  OUT-SB is non-zero for ONLY
   those prefixes selected for inclusion into the RIB-SB, what the
   corresponding attributes are.

   For example, if all external peers have NOT negotiated SBAB, those
   prefixes would have SB binary values of 10.  Each In-RIB-SB would
   have at most one instance.  And for each prefix, at most one
   In-RIB-SB would be selected as best, and have its corresponding
   OUT-SB set to binary value 10.

   This forward-chaining allows for processing of SB updates to
   determine whether withdrawals need to be flooded to peers, and if so,
   what SBAB attribute to apply to the withdrawals that are flooded.
   This flooding MAY be performed in parallel to normal BGP table update
   processing.

   For clarity, it should be pointed out that:

   o  The process for the step RIB-SB to RIB is "select prefixes marked
      'best'".

   o  The process for the step RIB-SB to out-RIB is also "select
      prefixes marked 'best'".

Dickson                  Expires August 4, 2008                [Page 12]

Internet-Draft         BGP Second-Best and Back-up         February 2008

   o  The process for the step RIB-SB to out-RIB-SB is the same as
      ordinary RIB to out-RIB, except for preservation of SB attributes
      (if any).

Dickson                  Expires August 4, 2008                [Page 13]

Internet-Draft         BGP Second-Best and Back-up         February 2008

5.  Security Considerations

   No additional security considerations beyond those already present in
   BGP are introduced.

Dickson                  Expires August 4, 2008                [Page 14]

Internet-Draft         BGP Second-Best and Back-up         February 2008

6.  IANA Considerations

   IANA will need to assign new code points for BGP Capabilities for
   USE_SECOND_BEST.  IANA will need to assign new code points for BGP
   Attribute Types for SECOND_BEST.

Dickson                  Expires August 4, 2008                [Page 15]

Internet-Draft         BGP Second-Best and Back-up         February 2008

7.  Acknowledgements

   The author wishes to acknowledge the helpful guidance of Joe Abley,
   Tony Li, and Yakhov Rehkter.  The author thanks the following for
   feedback during the review and revision process: Joel M. Halpern,
   Tony Li.  The author also wishes to acknowledge the insight gained
   from his Scottish Deerhound, Skylar, winning a Reserve Best-in-Show.
   (The selection method of "second best" comes from the Reserve system
   used at the group and best-in-show levels of dog shows).

Dickson                  Expires August 4, 2008                [Page 16]

Internet-Draft         BGP Second-Best and Back-up         February 2008

8.  References

8.1.  Normative References

   [1]  McPherson, D., Gill, V., Walton, D., and A. Retana, "Border
        Gateway Protocol (BGP) Persistent Route Oscillation Condition",
        RFC 3345, August 2002.

   [2]  Rekhter, Y., Li, T., and S. Hares, "A Border Gateway Protocol 4
        (BGP-4)", RFC 4271, January 2006.

   [3]  Chen, E. and S. Sangli, "Avoid BGP Best Path Transitions from
        One External to Another", RFC 5004, September 2007.

8.2.  Informative References

   [4]  Bradner, S., "Key words for use in RFCs to Indicate Requirement
        Levels", BCP 14, RFC 2119, March 1997.

Dickson                  Expires August 4, 2008                [Page 17]

Internet-Draft         BGP Second-Best and Back-up         February 2008

Appendix A.  Path-Hunting Examples

   (These will be included in a subsequent version of this ID.)

Dickson                  Expires August 4, 2008                [Page 18]

Internet-Draft         BGP Second-Best and Back-up         February 2008

Appendix B.  Persistent Oscillation Examples

   Consider the example in Figure 5 where o R1, R2, R3, R4, and R5
   belong to one AS. o R1 is a route reflector with R2 and R3 as its
   clients. o R4 is a route reflector with R5 as its client. o The IGP
   metrics are as listed. o External paths (a), (b), and (c) are as
   described in Figure 6.

   +----+      1      +----+
   | R1 |-------------| R4 |
   +----+             +----+
    |  \                |
    |   \               |
   3|    \ 2            | 6
    |     \             |
    |      \            |
   +----+  +----+     +----+
   | R2 |  | R3 |     | R5 |
   +----+  +----+     +----+
    |        |          |
   (a)      (b)        (c)

                                 Figure 5

   Path    AS_PATH MED
    a       1 3    10
    b       2 3     1
    c       2 3     0

                                 Figure 6

   With the addition of "second best", we have:

   R1 has the following:

   Path    AS_PATH MED IGP-metric
    a       1 3    10   3 (received:best) (best)
    b       2 3     1   2 (received:best)
    c       2 3     0   7 (received:best) (second_best - not sent)

   R4 has the following:

   Path    AS_PATH MED IGP-metric
    a       1 3    10   4 (received:best) (best - not sent)
    c       2 3     0   6 (received: best) (second_best)

Dickson                  Expires August 4, 2008                [Page 19]

Internet-Draft         BGP Second-Best and Back-up         February 2008

   This results in R1 having:

  Path    AS_PATH MED IGP-metric
   a       1 3    10   3 (received:best) (best)
   b       2 3     1   2 (received:best)
   c       2 3     0   7 (received:second_best) (second_best - not sent)

   By including the second_best in the best path calculation, the
   persistent oscillation problem is resolved.

Dickson                  Expires August 4, 2008                [Page 20]

Internet-Draft         BGP Second-Best and Back-up         February 2008

Author's Address

   Brian Dickson
   Afilias Canada, Inc
   4141 Yonge St,
   Suite 204
   North York, ON  M2P 2A8
   Canada

   Email: brian.peter.dickson <at> gmail.com
   URI:   www.afilias.info

Dickson                  Expires August 4, 2008                [Page 21]

Internet-Draft         BGP Second-Best and Back-up         February 2008

Full Copyright Statement

   Copyright (C) The IETF Trust (2008).

   This document is subject to the rights, licenses and restrictions
   contained in BCP 78, and except as set forth therein, the authors
   retain all their rights.

   This document and the information contained herein are provided on an
   "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS
   OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND
   THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS
   OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF
   THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED
   WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Intellectual Property

   The IETF takes no position regarding the validity or scope of any
   Intellectual Property Rights or other rights that might be claimed to
   pertain to the implementation or use of the technology described in
   this document or the extent to which any license under such rights
   might or might not be available; nor does it represent that it has
   made any independent effort to identify any such rights.  Information
   on the procedures with respect to rights in RFC documents can be
   found in BCP 78 and BCP 79.

   Copies of IPR disclosures made to the IETF Secretariat and any
   assurances of licenses to be made available, or the result of an
   attempt made to obtain a general license or permission for the use of
   such proprietary rights by implementers or users of this
   specification can be obtained from the IETF on-line IPR repository at
   http://www.ietf.org/ipr.

   The IETF invites any interested party to bring to its attention any
   copyrights, patents or patent applications, or other proprietary
   rights that may cover technology that may be required to implement
   this standard.  Please address the information to the IETF at
   ietf-ipr <at> ietf.org.

Acknowledgment

   Funding for the RFC Editor function is provided by the IETF
   Administrative Support Activity (IASA).

Dickson                  Expires August 4, 2008                [Page 22]

_______________________________________________
Idr mailing list
Idr <at> ietf.org
https://www.ietf.org/mailman/listinfo/idr
Samita Chakrabarti | 6 Mar 2008 01:53
Favicon

Re: comments on draft-chakrabarti-idr-rfc4893-mod-00.txt


Hi Enke,

Below is my response in-line. 

>> >
>>
>> > else needs to be said?
>>
>> */ [SC>] The issue-1 describes the need for clarification for
>> implementation on OPEN message processing when AS number is present in
>> both capability message and in "MY AS Number" fields./*
>>
>
>The text in the document is very specific and clear.
>
[SC>] 
         It is specific about putting AS_TRANS in "My ASN" field when 4-byte
unmappable AS number is present.  But I am talking about processing OPEN on
receive side. Where is the text on that in RFC 4893?

>> */ /*
>>
>> */ IMO, The document should clarify that AS Number present in
>> Capability message is used for the AS number and "My AS number" field
>> is ignored. From RFC 4893, it is assumed that nodes with
>> 4byte-mappable 2-byte AS number can advertise capability (AS4_PATH) as
>> well as nodes with unmappable 4-byte AS numbers. In case of
>> un-mappable 4-byte AS numbers,the "My AS number" field is AS_TRANS./*
>>
>
>Again, the document is clear about using AS_TRANS in the "My AS number"
>field:
>
>   To represent 4-octet AS numbers (which are not mapped from 2-octets)
>   as 2-octet AS numbers in the AS path information encoded with 2-octet
>   AS numbers, this document reserves a 2-octet AS number.  We denote
>   this special AS number as AS_TRANS for ease of description in the
>   rest of this specification.  This AS number is also placed in the "My
>   Autonomous System" field of the OPEN message originated by a NEW BGP
>   speaker, if the speaker does not have a (globally unique) 2-octet AS
>   number.
>
>
>> */ /*
>>
>> */ /*
>>
[SC>]   This talks about send side. But, I am talking about text on receipt.

>> >
>>
>> >-) On Issue - 2:
>>
>> >
>>
>> > The issues with having islands of the same AS number are well
>> understood.
>>
>> > Please see "RFC 2270: Using a Dedicated AS for Sites Homed to a
>>
>> >Single Provider".
>>
>> >
>>
>> > We could add the reference when there is ever sufficient reason to
>>
>> >update the document
>>
>> > in the future.
>>
>> */ [SC>] Yes. It should be updated. IMHO, it might be a good idea to
>> clarify and update the document now before the deployment starts and
>> face interoperability issues. Are there many implementations in
>> AS4_PATH that have been interop tested ?/*
>>
>
>Just FYI - there have been multiple implementations and deployments
>since 2004.

[SC>] Any pointers on deployment?

>
>In terms of transition, please see Section 6. It assumes a reasonable
>way of transition:
>
>   To simplify transition, this document assumes that an Autonomous
>   System could start using a 4-octet AS number only after all the BGP
>   speakers within that Autonomous System have been upgraded to support
>   4-octet AS numbers.
>
[SC>] 
       Yes. I saw that.  
>
>> */ /*
>> */ Con: Looking at AS_PATH or AS4_PATH info, one can not tell if the
>> transition has already happened in the Internet. But how important is
>> it to know this way? Was this a design goal ? /*
>>
>>
>
>This is a moot point at this time. After years of implementation and
>deployment, there can not be any change to the update generation.
>Besides what you proposed is something that was discussed and not
>pursued due to the major flaw I pointed out.
[SC>]
       If the idea on draft-chakrabarti was discussed before and not
adopted, then I'll not push this again. But the current solution has
weaknesses too that I pointed out in my last email.

>
>>
>> */ [SC>] I talked with a few people at NANOG who experienced that
>> AS_TRANS value was used often as AS number in the current BGP
>> deployment. We can not expect that every operator or administrator is
>> fully aware of reserved AS_TRANS value for this special purpose. Thus
>> the notification message from NEW BGP speaker to OLD BGP speaker who
>> uses AS_TRANS value as "My AS number" field in the OPEN message is
>> very helpful during transition./*
>>
>
>Sorry to say that you might be misinformed of the use of AS 23456. In
>the public internet, one must use an AS number allocated by IANA.
[SC>] 
[SC>]  I am well aware of IANA allocated AS number usage. That's not the
point.

>Otherwise there are consequences, including partial connectivities ....
[SC>] 
[SC>]  I agree. That's why a robust protocol takes care of sending some
error messages in some situations.

>
>The allocation and management of AS numbers are outside the scope of the
>document.
>
>> */ /*
>>
>> */ /*
>>
>> */ [SC>] besides, the transition section of RFC4893 should also
>> mention the order of AS number transition of PE/CE routers.Some folks
>> think that PE routers should be transitioned first (if not at the same
>> time) in order to take care of AS override function at the egress
>> path. [ it is not part of my draft yet ]/*
>>
>
>Please see Section 6 (Transition) of the document. The document assumes
>(probably should recommends) that the provider networks are upgraded
>first before connecting non-mappable 4byte AS peers.
>

[SC>]  I cannot see any specific words on 'provider networks' in section 6.

Please note that I am not trying to find flaws in RFC 4893, but requesting
some clarification on an important document that will be used for AS 4byte
transition. Clarity in the specification saves many interop issues and bugs
in the implementation, especially when something goes wrong in the Internet
domain routing, it can cause potential outage in the service. That's all.

Thanks,
-Samita

_______________________________________________
Idr mailing list
Idr <at> ietf.org
https://www.ietf.org/mailman/listinfo/idr

Samita Chakrabarti | 6 Mar 2008 02:11
Favicon

Re: comments on draft-chakrabarti-idr-rfc4893-mod-00.txt


Hi Geoff,

>>
>>    When transition is over, a BGP speaker can use AS4_PATH only. It is a
>> simple concept and no overloading.
>
>But there is no way for any individual BGP speak to know when transition
>is "over". I have to agree with Enke in observing that the proposal is
>indeed backwards.
>
>
[SC>]  I did not mean that a BGP speaker would know when the transition is
over. I mentioned that transition period would be long. During that period
both AS_PATH and AS4_PATH will exist and the transition border NEW BGP
speaker does not need to do 2byte<->4byte AS_PATH, AS_AGGREGATOR conversion.

But, Enke mentions that similar proposal was discussed before and not
adopted. In that case I'll not push for this particular proposal. But it is
true that RFC 4893 is overloading AS_PATH sometimes with 2byte and sometimes
with 4bytes values, the NEW speaker will have to carefully take care of this
situation. It could lead into error-prone implementations.

>>>-) On Issue - 4:
>>
>>>
>
>> */[SC>] I talked with a few people at NANOG who experienced that
>> AS_TRANS value was used often as AS number in the current BGP
>> deployment. We can not expect that every operator or administrator is
>> fully aware of reserved AS_TRANS value for this special purpose.
>
>I'm sorry, but this is a specious line of reasoning.
>
>Overall I see noting of use in this draft to warrant modification of
>RFC4893

[SC>] What is the objection against introducing a notification message to
the OLD BGP speaker from a NEW BGP speaker when it receives AS_TRANS in OPEN
message without AS4 capability?  This is quite useful for transition.

Regards,
-Samita

_______________________________________________
Idr mailing list
Idr <at> ietf.org
https://www.ietf.org/mailman/listinfo/idr

Geoff Huston | 6 Mar 2008 04:17
Favicon

Re: comments on draft-chakrabarti-idr-rfc4893-mod-00.txt


Samita Chakrabarti wrote:
> Hi Geoff,
> 
>>>    When transition is over, a BGP speaker can use AS4_PATH only. It is a
>>> simple concept and no overloading.
>> But there is no way for any individual BGP speak to know when transition
>> is "over". I have to agree with Enke in observing that the proposal is
>> indeed backwards.
>>
>>
> [SC>]  I did not mean that a BGP speaker would know when the transition is
> over. I mentioned that transition period would be long. During that period
> both AS_PATH and AS4_PATH will exist and the transition border NEW BGP
> speaker does not need to do 2byte<->4byte AS_PATH, AS_AGGREGATOR conversion.

would you care to quantify the overhead of any such conversion?

average AS path length = 3.5

Number of updates per second appears to avers around 3 to 4

I'm sorry but I simply can't see what you think you are saving here

> 
> But, Enke mentions that similar proposal was discussed before and not
> adopted. In that case I'll not push for this particular proposal. But it is
> true that RFC 4893 is overloading AS_PATH sometimes with 2byte and sometimes
> with 4bytes values, the NEW speaker will have to carefully take care of this
> situation. It could lead into error-prone implementations.

Yes, this was discussed about 3 years ago as well that the AS Path does 
not implicitly say whether the elements within the path are 2 or 4 
bytes. The consensus outcome of the WG was to run with what is in 
RFC4893. I think you you want to dredge over this ground again you are 
going to have to come up with hard facts, rather than another rerun of 
supposition and opinion of what "could" happen.

>>>> -) On Issue - 4:
>>> */[SC>] I talked with a few people at NANOG who experienced that
>>> AS_TRANS value was used often as AS number in the current BGP
>>> deployment. We can not expect that every operator or administrator is
>>> fully aware of reserved AS_TRANS value for this special purpose.
>> I'm sorry, but this is a specious line of reasoning.
>>
>> Overall I see noting of use in this draft to warrant modification of
>> RFC4893
> 
> [SC>] What is the objection against introducing a notification message to
> the OLD BGP speaker from a NEW BGP speaker when it receives AS_TRANS in OPEN
> message without AS4 capability?  This is quite useful for transition.

Lets see now what you are saying:

I configure to peer with AS 1. In the session the peer sends me an open 
message with a My AS number of 23456 without the capability to do 32 bit 
AS numbers - i.e. my peer is saying that it really is AS 23456. I 
terminate the session and spit out the diag that this is not the 
neighbor I expect to talk to.

Now the only way this will work is that I configure AS 23456 as my 
neighbor and you assert that you are AS23456. Fine.

Why would this happen ?

a) I am an OLD speaker and you have a non-mappable 32 bit AS number and 
you are a NEW BGP speaker. Well in this case this is precisely what is 
MEANT to happen

b) You and I agree that you really want to be AS23456. Well aside from 
the observation that you and I are demonstrating some level of ignorance 
about Standards and the IANA registry, no actual harm is being done.

So what is the problem you think that you are trying to solve here?

Hint: the AS Path is strictly speaking a loop detection mechanism and a 
comparison mechanism.

In what way would either loop detection or path comparison be impaired 
in the latter case?

Again, could we move away form the suppositions here and get to the hard 
facts?

regards,

      Geoff

_______________________________________________
Idr mailing list
Idr <at> ietf.org
https://www.ietf.org/mailman/listinfo/idr

Samita Chakrabarti | 8 Mar 2008 03:53
Favicon

Re: comments on draft-chakrabarti-idr-rfc4893-mod-00.txt

Please  see below in-line.

>> [SC>]  I did not mean that a BGP speaker would know when the transition
>is
>> over. I mentioned that transition period would be long. During that
>period
>> both AS_PATH and AS4_PATH will exist and the transition border NEW BGP
>> speaker does not need to do 2byte<->4byte AS_PATH, AS_AGGREGATOR
>conversion.
>
>would you care to quantify the overhead of any such conversion?
>
>average AS path length = 3.5
>
>Number of updates per second appears to avers around 3 to 4
[SC>] 
[SC>]  If there is 50+ peering connection then ~200 updates/sec for the
control-plane traffic. But one can argue that this is not worth worrying
about.

>
>I'm sorry but I simply can't see what you think you are saving here
>
>

[SC>]  My concern was more on overloading the AS_PATH data sometimes with
2-bytes and sometimes with 4-bytes values; it is error-prone from
programming point of view. 

But as I noted below that I understand the case is closed now. No point
arguing now.

>>
>> But, Enke mentions that similar proposal was discussed before and not
>> adopted. In that case I'll not push for this particular proposal. But it
>is
>> true that RFC 4893 is overloading AS_PATH sometimes with 2byte and
>sometimes
>> with 4bytes values, the NEW speaker will have to carefully take care of
>this
>> situation. It could lead into error-prone implementations.
>
>Yes, this was discussed about 3 years ago as well that the AS Path does
>not implicitly say whether the elements within the path are 2 or 4
>bytes. The consensus outcome of the WG was to run with what is in
>RFC4893. I think you you want to dredge over this ground again you are
>going to have to come up with hard facts, rather than another rerun of
>supposition and opinion of what "could" happen.
>
<snip>

>> [SC>] What is the objection against introducing a notification message to
>> the OLD BGP speaker from a NEW BGP speaker when it receives AS_TRANS in
>OPEN
>> message without AS4 capability?  This is quite useful for transition.
>
>Lets see now what you are saying:
>
>I configure to peer with AS 1. In the session the peer sends me an open
>message with a My AS number of 23456 without the capability to do 32 bit
>AS numbers - i.e. my peer is saying that it really is AS 23456. I
>terminate the session and spit out the diag that this is not the
>neighbor I expect to talk to.
>
>Now the only way this will work is that I configure AS 23456 as my
>neighbor and you assert that you are AS23456. Fine.
>
>Why would this happen ?
>
>a) I am an OLD speaker and you have a non-mappable 32 bit AS number and
>you are a NEW BGP speaker. Well in this case this is precisely what is
>MEANT to happen
>
>b) You and I agree that you really want to be AS23456. Well aside from
>the observation that you and I are demonstrating some level of ignorance
>about Standards and the IANA registry, no actual harm is being done.
>
[SC>]  The ignorance of standards and specification in the field is the real
concern - I am not trying to address any algorithm issue here.

>So what is the problem you think that you are trying to solve here?
[SC>] 

[SC>] Human error or inconsistency of the implementation due to lack of
clarity in the specification - is the problem I am trying to address here.

Assuming your example, above:

  R1> router bgp 77777
    Router> neighbor 3.3.3.3 remote-as 23456  <---- The correct new BGP
implementation should not allow this command to succeed. But it is possible
because the RFC explicitly does not prohibit that, as you mentioned.

On R2> router bgp 23456  [ OLD BGP and they have been doing this incorrectly
due to ignorance ]

OLD BGP sends OPEN message with My ASN = 23456, but there is no AS4
capability.
At this point, the NEW BGP should reject the OPEN and does not peer.
RFC4271 provides Notification type 2 subtype 2 (bad AS); if new BGP
implementation could use that NOTIFICATION before closing TCP, the debug
info could help the poor IT person who does not understand the details we
discuss here in IETF.

Here is a reference based on some real experience from the field about
common mistakes in peering configuration; it might look silly but it can
save customer calls($$) if the debug message shows some hint.
http://www.nanog.org/mtg-0802/presentations/PSmith_BGP.pdf - checkout slide
30

Thanks,
-Samita

_______________________________________________
Idr mailing list
Idr <at> ietf.org
https://www.ietf.org/mailman/listinfo/idr

Vishwas Manral | 8 Mar 2008 03:17
Picon

Hold Negotiation

Hi Sue/ Yakov,

I found an ambiguity in the Hold Time negotiation for the value of 0 received.

The idea of negotiation of the Hold time is that if we will use the
smaller of the two values of Hold time to be the actual Hold Time. A
value of 0, actually means the Maximum Hold time. So in case we get a
value of 0 from the neighbor, should we use the value of Hold time as
0(a smaller numeric value - but a bigger  value as such) or should we
use the value of (x > 0) as it is logically lesser than the value of 0
Hold time.

I would think we want the latter. Do let me know what you think?

Thanks,
Vishwas
_______________________________________________
Idr mailing list
Idr <at> ietf.org
https://www.ietf.org/mailman/listinfo/idr


Gmane