BBC does IPv6 ;) (Was: large multi-site enterprises and PI

Sat Nov 27 19:08:20 UTC 2004

On 27-nov-04, at 19:17, Paul Vixie wrote:

> i was waiting and watching and looking and hoping for this.  now i 
> have it.

Glad that I could oblige...

>> ... We have 128 bits, so we should make good use of them.  One way to
>> do this is to make all subnets and 99% of end-user assignements the
>> same size.  Yes, this wastes bits, but the bits are there anyway so 
>> not
>> wasting them really doesn't buy you anything at this point.

> this demonstrates the same unifying/universal error that fred was on 
> about,
> which is that some people doing resource planning making assumptions 
> about
> what the other people affected by those plans actually want or need.

So what's the alternative that you propose?

I really don't see the problem. The policy says: give a /128 to 
everyone who only needs a single address, a /64 to everyone who only 
needs a single subnet, a /48 to everyone who needs  1 < subnets < 64k 
and whatever is required to anyone who needs more than 64k subnets. If 
for some strange reason you want something different, like a /62, 
what's the harm in getting a /48 instead? There are 35184372088832 of 
those available in the global unicast space, so wasting the bits isn't 
an issue. However, the administrativia required to give out different 
sized blocks to different customers IS harmful for ISPs.

But it's useless to discuss this until you specify an alternative.

> the short version of my rebuttal is: "those are not your bits to 
> waste."

They are if my ISP assigns them to me.  :-)

>> All I hear is how this company or that enterprise "should qualify" for
>> PI space.  What I don't hear is what's going to happen when the
>> routing tables grow too large, or how to prevent this.  I think just
>> about anyone "should qualify", but ONLY if there is some form of
>> aggregation possible.  PI in IPv6 without aggregation would be a
>> bigger mistake than all other IPv6 mistakes so far.

[...]

> | my recommendation that we face
> | the reality that PI is an important thing (unless we want to 
> replicate
> | the v4 NAT mess).  As such, I'd much rather see us develop sane PI
> | policy than continue down the present road.

I'm all for a sane PI policy. However, the current argument is like 
this: someone is standing on top of a burning building. Some people are 
shouting: "Don't jump, it's too high, you'll be killed!" and others are 
shouting: "Jump! If you don't, you'll be fried!"

I fully agree that with the current state of technology, IPv6 without 
PI isn't a good deal for larger organizations. However, changing the 
IPv6 policies so that people can have PI space is very dangerous, as 
some natural limits on the number of PI or PI like blocks that exist in 
IPv4 don't exist in IPv6, and we need IPv6 to be around for a long time 
to come.

Fortunately, we don't have to choose between a rock and a hard place: 
we can change the technology so that the drawbacks of both PA and PI 
are reduced. For PA this is renumbering and multiaddress multihoming, 
for PI this would be building in the potential for aggregation.

> second, let me add, "and it's not your routing table, either."

I have no idea what this means.

> to make ipv6 take off we'll either have to grind down the folks who 
> don't
> want to be locked in (and have their downstreams locked in) to a single
> upstream; or we'll have to insert rapid renumbering into a design that
> makes no allowance for it; or we'll have to let PI happen in ipv6 as it
> has in ipv4 -- through careful equilibrium;

There is no such thing. Done a "show ip bgp" lately? The v4 routing 
table is a huge mess and it's getting worse by the week. We need to do 
better and we can do better in IPv6.

> or we'll have to let NAT happen in ipv4 as it has in ipv6.

Why do v6 if you're going to NAT anyway???

> the delicious thing about those prescriptions is that there is no "we"
> and it's not up to "us".  what will actually happen is something we can
> predict before, and describe after, but not actually control.

None of us individually can control the big picture. But if enough 
people decide that's something is a bad idea and don't cooperate, it's 
not going to happen. That's why I'm not afraid of ULAs being routed. IP 
has never been about shoving stuff down people's throats, but about 
everyone building the best network that they can within the limitations 
of clue, money and technology and thereby creating the best possible 
internet.

> i predict
> that a bunch of ivory tower propeller heads will block everything they
> think is impure and that the market will have to decide on "dual-stack
> forever with NAT on both stacks."

I predict that most people will realize that IPv6 needs to be better 
than IPv4 in some important aspects to be a valid successor to IPv4, 
and it's better to wait a bit longer for something good.