[e2e] the evolution of deployability

[RJ Atkinson]

On Tuesday, Dec 17, 2002, at 13:35 America/Montreal, Henning 
Schulzrinne wrote:
> I also suspect that the major vendors of NAT boxes are neither C nor 
> M, but rather Linksys, DLink, Netgear and similar consumer-oriented 
> vendors. The primary motivation for NATs is not selling boxes, but the 
> exorbitant cost of IPv4 addresses. Even for a business DSL, last time 
> I looked, a single IPv4 address costs $5/month. My suspicion is that 
> ISPs use the IPv4 scarcity to reduce bandwidth usage by making 
> deploying web servers, Shoutcast and P2P servers difficult. They have 
> an active interest in not making IPv6 happen, much more than M or C, 
> which would presumably be more than happy if customers had to upgrade 
> operating systems or routers. (Conspiracy-minded folks could imagine 
> nothing better for M than if the Internet were to convert to IPv6 on 
> January 1 and only Windows XP supported IPv6.)

Charging per-month for additional (i.e. 1 free, but the rest have a 
recurring charge)
for residential US cable modem service was basically insisted upon by 
ARIN.
Many cable network ISPs make no profit on that option, having it only 
to keep
the address block regulators happy.  An unhappy address block regulator 
could
keep an ISP from adding subscribers, so it is a pretty effective tool.

ISP disinterest in IPv6 most is due to (1) increased operational costs 
of supporting
2 network-layer protocols instead of 1 network-layer protocol, (2) lack 
of
end-user customer demand, and (3) lack of any ISP operational benefit 
to deploying
the new protocol.  (3) could be fixed if IPv6 had a more scalable 
routing
architecture; right now, IPv6 has the same routing architecture as 
IPv4, so
no particular ISP benefit to IPv6 deployment.

Ran
rja at extremenetworks.com