IPv6 - a kludge wrapped in an old gym sock wrapped in week-old fish-wrap

Mike O'Dell mo at 131.ccr.org
Thu Jan 26 14:08:02 CST 2012

I wrote this as a rejoinder to the ED article but in line with my
New Years Resolution to "Kick fewer anthills," i didn't post it in
the COMMENTS section.

But since Andre' picked-open the wound.....


Unfortunately, other than the apparent size of addresses (which
are nowhere as large as they first appear), there is in fact
very little difference be tween IPv4 and IPv6. Yes, some of
the packet fields got moved around and the names changed to
confuse the innocent, but all of the fundamental architectural
flaws present in IPv4 are still there in IPv6. Worse, the
impact of the uncorrected architectural errors is amplified.

As for the "larger addresses" frequently cited as the greatest
asset of IPv6, having LOTS MORE of a problem you already cannot
solve does not improve matters. Addressing in IPv4 was the
triumph of duct tape over inadequate design. The larger addresses
in IPv6 thoughtfully provides for much longer, wider rolls of
duct tape but with no reduction in the necessity of its use.

To wit...

"IP Mobility" is said to make IPv6 Just The Thing for mobile
endpoints like smartphones. Unfortunately, it has been known
for a decade that the IP Mobility scheme for IPv6 is an extremely
problematic solution for the wrong problem. This results from
multiple IPv4 design failures interacting badly. Providing
application sessions which survive dynamic transport protocol
fail-over is readily accomplished with a modest session protocol.
Instead, IPv6 provides an enormously-complex edifice which
dynamically disguises the identity of network endpoints in an
attempt to trick the transport protocol into not failing. This
approach strikes directly at the heart of the most fundamental,
revered tenet of the design in the Internet - the End-to-End
Principle. How the IETF could do such a thing is puzzling
indeed. (Actually, it's not puzzling in the slightest, it's
just deeply embarrassing.)

It is clear IPv6 will be widely adopted in spite of its failings
because Them What Say So have said "Something Must Be Done!"
And so it shall.

The best comment on this argument was offered by Rob Pike of
Bell Labs rega rding other dubious-but-popular software:
"Something can fill a vacuum an d still suck."

Just so.

More information about the Tacos mailing list