Instead of trying to reply each email I will attempt to summarize and 
respond to the comments in the last set of email.

1) Requirement level of SHOULD would make existing implementations 
non-compliant.

I believe that the definition of SHOULD (see RFC2119) allows for alternate 
behavior if there is a good reason for it.  I think this can cover both 
better algorithms for the selection of default routers and if the 
implementation predates this spec.  I will note that the intent of the this 
document is to change the default behavior of implementations.  That is way 
I think SHOULD is appropriate.  Also, I think this is independent of the 
level of the standard (Proposed, Draft, Standard).

2) Relationship of this change to Neighbor Discovery and Default Router 
Preferences draft.  Should they be combined?

The reason why the conclusion of the w.g. in Salt Lake City was to keep the 
documents separate is that this document is changing the base behavior of 
Neighbor Discovery when there is more than one default router.  The Default 
Router Preferences draft is an optional mechanism (see the draft).  It may 
not always be used and/or implemented.  It was agreed that the Router 
Preferences draft should also incorporate this behavior when there are more 
than one router of the same preference.

Note this draft is not replacing the router preferences draft.  Both are 
expected to go forward.  This draft only changes the case when there is 
more than one default routers.  It does not provide the optimum solution 
when there are multiple routers with different characteristics.  That case 
is covered by the router preferences draft.

3) General statement load-distribution is not predictable behavior and is 
undesirable (e.g., using a single router until it fails is better).

There is considerable current practice that disputes this.  The specific 
case that this draft addresses (host to router traffic) is very widely 
deployed using protocols such as VRRP and Cisco's HSRP.  The advantage of 
distributing the load over multiple routers is that it takes advantage of 
the resources that are available and insures that the backup is working 
when it is needed.  Backup systems that are not being used have a nasty 
habit of not working when they are needed.

Regards,
Bob



  

--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to