[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: The state of IPv6 multihoming development



On Tue, 22 Oct 2002, Shane Kerr wrote:

> A multi-PA multihoming solution can use several techniques to allow
> the destination host to determine the addresses used, especially if it
> is layer 4 or higher.  It seems like a lot of of end-sites would be
> happy with redundancy first, speed second, and least-cost routing as
> icing on the cake.

Enterprise end-sites want equivalent functionality to IPv4.  It also has
to be manageable, such that the cost of managing the multi-homing solution
is equal or less than that of managing multi-homing in IPv4 today.  They
need to be able to trust the infrastructure enough to make their AAAA RR
part of the records for www.<domain>.com.

Given the business environment of the IPv4 Internet, I don't think you can
trade off speed for redundancy.  Multi-homing in IPv4 today delivers both.
We need to deliver both.

> Do you think enterprises would you reject out of hand a solution that
> uses DNS-style RTT optimisation for destination address selection?

When the host selects the path by virtue of address selection, such that
the routing infrastructure can see no alternate paths to the destination,
then enterprise network operations teams have no capability to manage
traffic policy.  I believe the majority of enterprise network operations
teams would have an issue with that.

We should keep traffic routing within the network layer, IMO, or at least
give visibility to the network layer of alternate paths that can be taken
for a variety of policy reasons.

> > BGP isn't perfect either, but it's much better than how many bits a
> > particular address has in common with another.  Managing n+1
> > prefixes per subnet where n is number of providers serving a site is
> > a nightmare.  The list of problems goes on...
>
> Maybe you can point me to a URL of the specific draft you're
> addressing here (I only see the hoary requirements document at the
> IETF site).

draft-ietf-ipngwg-default-addr-select-06.txt specifies the use of the pair
of addresses with most common bits (left-most) in a set of source and
destination address pairs.

The IPv6 specifications for aggregatable address architecture offer that
for multihoming, each segment could have multiple IP subnets from
different PA-assigned /48's for an end-user.  The host then uses the
selection mechanism in default-addr-select to select the source and
destination addresses to use.  An enterprise would have to assign, per
segment, a /64 from PA space from each provider plus perhaps a site-local
/64 as well, and manage that list of prefixes across the entire enterprise
network.

/cah

---
Craig A. Huegen, Chief Network Architect      C i s c o  S y s t e m s
IT Transport, Network Technology & Design           ||        ||
Cisco Systems, Inc., 400 East Tasman Drive          ||        ||
San Jose, CA  95134, (408) 526-8104                ||||      ||||
email: chuegen@cisco.com       CCIE #2100      ..:||||||:..:||||||:..