Re: Market forces

[Mark Smith <[e-mail]>]

On Mon, 9 Nov 2020, 23:01 Ted Lemon, <[e-mail]> wrote:

> On Nov 9, 2020, at 5:06 AM, Simon Hobson <[e-mail]> wrote:
>
> Have you heard of the expression "don't shoot the messenger" ?
> People are telling you what they observe. Right or wrong, there **ARE**
> ISPs out there where "bottom line" is the only thing they care about, and
> customer experience come last on the list of priorities.
>
>
> Simon, the thing you’re glossing over here, which I actually don’t hear
> anybody talking about, is that there has thus far been no market pressure
> that could punish ISPs that hand out /64s. What’s the bad outcome for the
> end user in practice? They can’t subnet their home network? Well, they
> don’t, do they? HNCP has a market penetration of 0% right now.
>

I think the current and existing need that wouldn't be being satisfied with
a single /64 are things like a guest WIFI SSID in the home, or perhaps a
kids SSID where Internet access policy can be applied.

With IPv4, the current home internal address space is effectively the
entire RFC1918 space behind the NAT implementation.

The common exception is when an employer's RFC1918 space needs to be
excluded so the employee can VPN in from home without traffic instead going
to the local instance of that address space.

It seems fairly common for corporates to use 10/8, or sometimes 172.16/12,
and common for IPv4 CPE to use 192.168/16, naturally avoiding this address
space duplication issue.

So limiting to 192.168/16, with the /24 common subnet length, IPv4 users
have 256 /24s for their CPE to dole out for these purposes, equivalent to
an IPv6 /56.

Regards,
Mark.



> We can reasonably expect this to change in the next year or two. Several
> initiatives are underway at the moment that are likely to suffer from the
> absence of a narrower allocation: specifically CHIP and Thread, both of
> which rely on 802.15.4.6lowpan networks. It’s reasonable to expect that
> these technologies will start to be deployed in homes in significant
> numbers in the near future.
>
> At present, when such technologies are deployed, they will be forced to
> rely on IPv4 for cloud service reachability for ISPs that don’t provide an
> allocation narrower than /64. Since these networks are v6-only, this will
> be done using NAT64. And this will work fine, except that the ISP is now
> stuck giving the customer an IPv4 address.
>
> What is less scarce than an IPv4 address? Why, a /60 or a /56. These are
> 2^28 or 2^24 less scarce than IPv4 addresses, respectively.
>
> So all this fuss about how to solve the problem of ISPs only giving out
> /64s is genuinely premature. Market forces will change their behavior. It’s
> just a matter of time.
>
> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> [e-mail]
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------
>