Jon Crowcroft writes:
| anyhow, i think the old 8+8 v6 scheme would have sorted this out just
| dandily....and i dont understand the vitriol people our on it -
I don't understand alot of the vitriol, but I personally am concerned
that 8 bytes is insufficient. If there was ever a time-space trade-off
that should favour more space, it's addresses in future networks that
require packet order to be maintained (at least within individual flows).
Clearly there are other bits of a size-large Internet which will
be many orders of magnitude slower in terms of pps and bps, exposing the
fact that no given space-versus-compute-time tradeoff in packet
headers will satisfy everyone.
Result: RFC 1707 redux.
The VLA arguments, with "tunable" address lengths (to some maximum)
is an interesting compromise with people who have trouble coping
with arbitrary address-length variability, but I think CATNIP's
scheme is just as good, allowing for "tunable" networks, some of which
may choose to find an appropriate per-packet space-for-compute-time
tradeoff.
The vitriol that will be poured on this is from reactionaries
who seek to preserve the indistinction between who and where,
and who seek genetic purity of the network layer (i.e., the address
doesn't change at any border, implying the same protocol in use
everywhere).
My bet is that at least some of the 8+8 opponents realize that
a move away from current IPv6 + strict CIDR + single namespace
on either of the latter two vectors inevitably leads to a system which
encourages fractional and ongoing obsolescence of IPv6 itself,
perhaps even before it's widely deployed.
Sean.