At 6:47 PM -0700 6/3/01, Randy Bush wrote:
> > "There is little experience with widespread, arbitrary use of internet
> > anycast addresses
>
>actually, there is a fair bit of experience with it.
Randy,
There wasn't when that sentence was written. (That's the hazard of
including such sentences in a long-lived document -- another thing
for the RFC editor to watch for?)
>as far as i can tell, the base problem here is that folk who have had no
>experience with v4 anycast service deployment know in theory or in vitro
>that there can be problems in some circumstances
This is a case where we don't need to run the program to demonstrate that
there's a bug in it.
>...
>
>unfortunately, the result has been outlawing a whole bunch of stuff which
>can not exhibit this kind of problem, namely udp based anycast services,
>and outlawing tcp services which may not be as susceptable to the problem
>about which they seem to be worried.
The restriction doesn't outlaw such services; it just requires that
those services satisfy the restriction, which may require a simple
change to the service in some cases (those that weren't fixed years
ago to accommodate multicast queries).
>[0] - in fact, in real deployment, long lived tcp to a v4 anycast seems
> more stable than even optimism would seem to warrant.
So the bug doesn't show up often in practice. That doesn't mean the bug
doesn't exist and won't have harmful consequences in those rare occasions
when it is triggered.
>research continues.
Research in finding and understanding the corner cases?
Steve
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
