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

Re: Again no multi6 at IETF#56




	Tony,

| > All of the pain and none of the gain? Any announcement of longer
| > prefixes is at the expense of routing scalability.
|
| Agreed. No question. But with 350 routes I am more worried
| over the
| lack of routes....


Once upon a time, not so long ago, the entire IPv4 Internet was only
1000
routes. Since then, I've been fighting exponential growth. Be careful
what you wish for.
I am well aware of this fact and I have always said that this is not the answer for any longer term. It's just that with only 350 routes people are already announcing much longer prefixes and are successful in suing them. This is just copy the routing operational practices from ISPs today.

|    > The cost of multihoming is falling.  Today, sitting at home on
|    > my couch, I can easily connect to my neighbor's WAP.  If we
|    > were to tweak things a bit, both of us would be multi-homed
|    > through the other.  In fact, the cost of connectivity is
|    > continuing to fall, so one might expect that the cost of
|    > multi-homing will fall too.
|
|    I don't question that. What makes me wonder is how many
|    home users will
|    actually want/need multihoming? How many SOHO will? How many
|    enterprises?


Historically, about 10% of all Internet sites are multi-homed.  I
see no reason for this not to continue.  As to homes, I can tell
you that my provider is certainly not so reliable that I am happy
being singly-homed.  And I'm not alone, I know of many others
with similar situations.
I am sure there are. But price is still a factor that you need to calculate with.


|    > If you subscribe to the view that businesses find the Web
|    > interface to be important, then it stands to reason that the
|    > interest in multihoming would continue to grow.
|
|    Agreed, the question is how fast?


Faster than we can possibly accommodate.  Our job is to provide the
technical leadership and be there with an architecture BEFORE the
masses show up.  Because by then, it's too late.

Agreed, still we need to try and guess when that will be. Mostly because that will set the agenda. I would be interested in seeing what the trend curves for the IPv6 routing table looks like from the people monitoring these. But I am not sure if Geoff Huston or Gert During for example are on this list.

|    Well, I think we need to move in steps for a number of reasons,
|    urgency, implementation lead times, and last to get
|    consensus. This
|    said, I think that we need to always keep in mind how we
|    will get from
|    A to B, and eventually to C.
|
|    But that is what I hope will be the first issue we could
|    get agreement
|    here on.


I agree that we need all of those things.  However, I don't believe
that we're in an undue rush to move forward.  Providing an interim
solution is simply going to turn that interim into a final solution,
and we will regret it.

Maybe. But coming to that conclusion and consensus is also an achievement we have yet to make.

- kurtis -