[afnog] Another case for RPKI ?

[Job Snijders]

Hi Bill,

On Wed, Nov 14, 2018 at 07:33:40PM +0100, Bill Woodcock wrote:
> Mbong has a point. If BGPv4 is the last version of BGP, we might as
> well write the Internet off as dead now. When the IETF stops working
> on new protocols, the Internet stops evolving. And for anything that
> doubles in size every ten and a half months to become static, to
> fossilize, is to die. 

We're now actually at something like BGP 4.2489 (I made up the point
number for illustrative purposes), the 2489 number will keep
incrementing as IDR and GROW continue their work. Because the Internet
keeps growing (and a large degree of backwards compatibility is
required) to keep it running, all current innovation in Internet context
is incremental. I don't see another way that is equally deployable.

> At the same time, it’s equally true that BGPv5 is “not just sitting
> around.”  It’ll take work. By people going to the IETF and doing that
> work. That’s the problem with the same old (and older and older)
> people showing up at the IETF every year: very little truly new work
> gets done. Just old graybeards layering patches over patches over
> patches on protocols we designed in our twenties. 

You paint a dark image which doesn't match my personal observations at
IETF 103. Its easy to shrug off IETF as 'old graybeards', but saying
such things doesn't necessarily make it true.

> So. Go to it.  Go to the IETF, ignore all the old dudes telling you
> they’ve seen and done it all before and that nothing you’re doing is
> necessary, and crank out  a BGPv5.  Do it for the network operators
> who will have to keep doubling the size of the network after us
> crotchety old dudes are safely dead. 

As I said, I look forward to the proposal. OP made two statements, one
reflected negatively on RPKI, the other pointed at a non-existing
solution. I'm interested in this problem space and would like to know
why such statements are made and what we can do to improve.

It's easy to handwave and say "we need a new BGP that will solve all
issues", but I'd like to see actual proposals. Until then we'll have to
keep working to make BGP-4 deployments the best we can, simply because
we have nothing else.

Kind regards,

Job