On Sun, 4 Apr 2010, Christopher Morrow wrote:

also, see previous 12 episodes of this conversation.. 1 /8 == ~3months in ARIN allocation timeframes.

Does a trade show really need 16M IPv4 addresses though? How many other /8's were assigned way back when IPv4 was being given out so freely that ARIN would laugh at if that org applied today for that /8? If we could recover them all, how many more years of IPv4 allocations would that buy us? ---------------------------------------------------------------------- Jon Lewis | I route Senior Network Engineer | therefore you are Atlantic Net | _________ http://www.lewis.org/~jlewis/pgp for PGP public key_________

On 5 Apr 2010, at 9:13, Jon Lewis wrote:

On Sun, 4 Apr 2010, Christopher Morrow wrote:

[...]

If we could recover them all, how many more years of IPv4 allocations would that buy us?

We allocate RIRs approximately one /8 per month. So you'd have to reclaim 12 /8s to extend the allocation pool by one year. Regards, Leo

Might want to double check you aren't filtering, as parts of 1/8 and 2/8 have been intermittently announced by RIR's in debogonizing efforts over the last few months. Routing wise, this really isn't different from the space being assigned - better to clear up any filtering and identify routing problems or renumbering efforts you may need now before the space gets allocated, probably later this year. In fact, parts of 2/8 are being announced right now for debogon-izing: route-views>sh ip bgp 2.0.0.0/8 longer-prefixes BGP table version is 2323163774, local router ID is 128.223.51.103 Status codes: s suppressed, d damped, h history, * valid, > best, i - internal, r RIB-failure, S Stale Origin codes: i - IGP, e - EGP, ? - incomplete Network Next Hop Metric LocPrf Weight Path * 2.0.0.0/16 194.85.102.33 0 3277 3267 30132 12654 I --Heather -----Original Message----- From: John Palmer (NANOG Acct) [mailto:nanog2@adns.net] Sent: Tuesday, April 06, 2010 7:37 PM To: nanog@nanog.org Subject: Re: interop show network (was: legacy /8) When do you think that 1/8, 2/8 and 50/8 will start showing up as live, assigned addresses. I don't see any of them coming in on my core routers yet. ----- Original Message ----- From: "Leo Vegoda" <leo.vegoda@icann.org> To: "Jon Lewis" <jlewis@lewis.org> Cc: <nanog@nanog.org> Sent: Monday, April 05, 2010 1:04 PM Subject: Re: interop show network (was: legacy /8) On 5 Apr 2010, at 9:13, Jon Lewis wrote:

On Sun, 4 Apr 2010, Christopher Morrow wrote:

[...]

If we could recover them all, how many more years of IPv4 allocations would that buy us?

We allocate RIRs approximately one /8 per month. So you'd have to reclaim 12 /8s to extend the allocation pool by one year. Regards, Leo

On Tue, Apr 06, 2010 at 12:01:47PM -0400, Robert E. Seastrom wrote:

Christopher Morrow <morrowc.lists@gmail.com> writes:

...

also, see previous 12 episodes of this conversation.. 1 /8 == ~3months in ARIN allocation timeframes.

1 /8 at global IANA free pool runout time (which is the only reasonable way to think about it...) will buy us about 24 days on a global consumption basis... assuming there isn't an end times land rush.

-r

so... just for grins, how do all those w/ bits leftover from their "overly generous" inital allocations (they forced me to take a /20 when all i really needed was a /28 multihomed) find partners who are willing to use the rest of those bits to get the delegation in question up to a comfortable 85% utilization? are you and Marty going to open up a matching service? co-op "match.com" or "e-harmony"? craigslist? --bill

On Wed, Apr 07, 2010 at 10:12:48AM -0400, Brandon Ross wrote:

On Wed, 7 Apr 2010, Eliot Lear wrote:

...

If v6 is even close to ready, wouldn't it be sad that this sort of testing isn't done at interop? Or is it just sad that v6 isn't so close to being ready? Or is it both?

The suggestion was to run a "v6 only network". Does anyone on the NANOG list believe that v6 is at all ready to be run without any v4 underpinnings and provide a real service to a customer base?

-- Brandon Ross

very - very close. if you have fewer than 50,000 nodes in your net, and its not topologically dense, then you -can- run a native IPv6 net w/o dual stack (save on the edge translator and the DNS (and DHCP - if you have the patches)) for all of them. I've done it - on several networks. --bill

The suggestion was to run a "v6 only network". Does anyone on the NANOG list believe that v6 is at all ready to be run without any v4 underpinnings and provide a real service to a customer base?

If you're an MPLS provider (as we are), the lack of IPv6 LDP is a major showstopper. Steinar Haug, Nethelp consulting, sthaug@nethelp.no

On Wed, 7 Apr 2010, Eliot Lear wrote:

I remember the days of Ron Natalie running around with a cherry picker in San Jose, and the whole point of the network being to test interoperability, so that things would and did break (and then we fixed them). If v6 is even close to ready, wouldn't it be sad that this sort of testing isn't done at interop? Or is it just sad that v6 isn't so close to being ready? Or is it both?

The lack of v6 readiness for a long time (and to some extent today) seems to have been locked in a vicious circle. Many users haven't been pushing vendors for v6 capabilities in their products (software and hardware) because they either didn't know about it, and/or didn't perceive it as important. OS developers seemed to be the most ahead of the curve on this, with usable v6 stacks available for most modern OSen for several years, and close to a decade in some cases. Many providers for a long time weren't implementing v6 because, while many knew it needed to happen, customers weren't pushing for it, and many network equipment vendors didn't have solid v6 implementations. Content providers would also fall into this bucket. Many vendors for a long time weren't making v6 development and support a priority because customers weren't pushing for it, so they didn't see a financial reason to do so. jms

From: Valdis.Kletnieks@vt.edu Date: Wed, 07 Apr 2010 12:28:39 -0400

On Wed, 07 Apr 2010 16:09:25 +0200, Eliot Lear said:

...

them). If v6 is even close to ready, wouldn't it be sad that this sort of testing isn't done at interop?

Interop long ago ceased being a interop shootout and became a 8x11 color glossy trade show. I think the last time any actual *testing* happened at Interop, the guys hooking up the network drops were wearing t-shirts that said "Yes, the subnet mask really *is* 255.255.252.0", and anybody who whined that their gear only supported octet-boundary subnets was told "And next year, it will be 255.255.250.0".

:)

Anybody got production gear that *still* doesn't do non-octet-boundary subnets?

Sort of. We discovered that Infinera DWDM gear's boot ROM OS is entirely classful. Not only most the subnet mask be on octet boundaries, but the boundary must match the address space used. (Well, you really can't specify a subnet mask. It just sets it based on the address.) That said, this is only the boot ROM system. Once you figure out how to make that work in your configuration, you load the real, running software which does CIDR and any subnet mask. So the impact is minimal, but it did shock me when I ran into it. -- R. Kevin Oberman, Network Engineer Energy Sciences Network (ESnet) Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab) E-mail: oberman@es.net Phone: +1 510 486-8634 Key fingerprint:059B 2DDF 031C 9BA3 14A4 EADA 927D EBB3 987B 3751