Re: Anycast support in draft-ietf-ipv6-addr-arch-v4-02.txt

[Joe Abley <jabley at isc.org>]

Hi Bob,

On 13 Apr 2005, at 17:13, Bob Hinden wrote:

>    Arbitrary use of Internet anycast addresses is not recommended.  
> There
>    are known complications and hazards when using them in their full
>    generality [ANYCST].  Specific usage guidelines are:
>
>       1) Anycast may be used for simple query response applications
>          (for example DNS) where all nodes serving the anycast
>          address will respond with the same information and the packets
>          are limited in size so path mtu discovery is not needed.
>
>       2) Anycast may be used for applications where anycast is used to
>          rendezvous with a server and subsequently learn a stable 
> unicast
>          address for further communication.
>
>       3) Except as described in 1) and 2) above an anycast address must
>          not be used as the source address of an IPv6 packet.
>
>       4) Except as described in 1) and 2) an anycast address must not 
> be
>          assigned to an IPv6 host, that is, it may be assigned to an 
> IPv6
>          router only.
>
> Comments and suggestions welcome.

I think the issue of when it is appropriate to use anycast addresses is 
not going to be reasonably summarised in five paragraphs, and I caution 
against attempting to do so. By way of illustration, the document 
draft-ietf-grow-anycast-00 runs to 20 pages. (That draft could use 
improvements, but it doesn't contain 19 pages of filler :-)

Might it be better to simply note that anycast is not a 
universally-appropriate technique for service distribution, and that 
caution is required? From some vantage points the text you proposed 
doesn't actually say much more than that (since most services can be 
described as "simple query response applications" when deconstructed). 
If details are required, a reference to the grow document would provide 
better guidance to the reader.

Specific points raised by the text above include:

1. There are services distributed using anycast in IPv4 which 
deliberately provide different information from different nodes, in 
order to provide topologically-sensitive service. It is not clear that 
this is unreasonable, or should be prohibited in IPv6, as your 
suggested item (1) above does.

2. There are examples of "simple query response applications" that 
perform very badly when distributed using anycast, including the DNS. 
The suitability of anycast lies both in the nature of the protocols 
concerned and also the nature of the network across which the service 
is being distributed.


Joe


--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6 at ietf.org
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------