Re: [MBONED] [Ext] Re: ipv4 multicast address ranges, actual usage.

[Dave Taht <[e-mail]>]

On Thu, Dec 19, 2019 at 9:29 AM Leo Vegoda <[e-mail]> wrote:
>
> On 12/18/19, 3:16 PM, "MBONED on behalf of Dave Taht" <[e-mail] on behalf of [e-mail]> wrote:
>
> [...]
>
> > Do you think anybody would notice if 225/8 - 231/8 became unicast? We
> > have scanned the source code of the world and come up empty here.

Thx everybody for your feedback.

>
> I think the amount of attention required would vary depending on the defined scope. If you want to expand the RFC 1918 address space, most networks can ignore the change unless they want to add support in their own administrative domain (unless you leak). If you want to change them into Global Unicast addresses then multiple, independent network operators need to agree to deploy new code in operational support systems as well as network elements.

All true. But regardless of ultimate designations of global unicast or
private rfc1918-like address spaces, it needs to be technically made
to work, and made routable. I do see a pretty big need for a larger
CGN space somewhere (and not that I get to pick anything!!!), I'd use
230/7 for that.

In addition to creating the patches and a testbed, I've given two
talks on this, the better one is marred by a clothing malfunction,
but: https://www.youtube.com/watch?v=92aNK3ftz6M

> I should note that redesignation for 1/8, 233/8, and 240/4 has been proposed before:
>
> https://tools.ietf.org/html/draft-hain-1918bis-01
> https://tools.ietf.org/html/draft-wilson-class-e-02

The class-e work was successful. Only windows of the major OSes did
not enable 240/4 as a viable address range 10 years ago. Adding
support for various routing protocols has been straightforward. I
sometimes think fondly about a portion of the internet entirely free
of windows born viruses and worms rather than trying to encourage them
to add support.

Linux finished the 240/4 fix (it worked in the "ip" command, not in
the "ifconfig" command) last year at around this time, 0/8 went in 6
months ago. Up next is this discussion of multicast spaces (which I
hope to add some new text to the drafty draft)

Had not the need for more space not been met successfully by CGNs back
then perhaps it would have moved forward in the standards.

>
> While those are not multicast address blocks, I think that the likely challenges are sufficiently similar to make it worth speaking with the authors of those drafts to understand why their proposals did not result in changes.

We did, and cited their work in our still drafty draft, here:

https://github.com/dtaht/unicast-extensions/blob/master/rfcs/draft-gilmore-taht-v4uniext.txt

I would love to collect more feedback and bug reports regarding that
before we finally gird ourselves up for the ietf gauntlet. As things
stand today, although I'd really like as much feedback as possible, I
do not see a problem (over a 10 year timespan) with turning
225/8-231/8 into unicast.

>
> Kind regards,
>
> Leo Vegoda
>


-- 
Make Music, Not War

Dave Täht
CTO, TekLibre, LLC
http://www.teklibre.com
Tel: [phone]