I'm wondering (and that shows that I have nothing better to do at 3:30am
on Monday...) how many people around here realize that the plain old
IPv4 - as widely implemented and specified in standard RFCs can be
easily used to connect pretty much arbitrary number (arbitrary means
2^256) of computers WITHOUT NETWORK ADDRESS TRANSLATION. Yes, you hear
me right.
And, no, it does not require any changes any in the global routing
infrastructure - as implemented now, and most OS kernels (those which
aren't broken-as-designed, grin) would do the trick just fine. None of
that dual-stack stupidity, and, of course, no chicken-and-egg problem if
the servers and gateways can be made to respect really old and
well-established standards.
DNS and most applications would need some (fairly trivial) updating,
though, to work properly with the extended addressing; and sysadmins
would need to do tweaks in their configs since some mythology-driven
"security" can get in the way. But they don't have to do that en mass
and all at once.
The most obvious solution to the non-problem of address space shortage
is the hardest to notice, ain't it?
This seems like either truly bizarre trolling, or the misguided idea of someone who's way too invested in IPv4 and hasn't made any necessary plans or steps to implement IPv6. To implement this -- which, to begin with, seems like a bad idea to me (and judging by Mr. Andrews' response, others) -- you'd have to overhaul software on many, many computers, routers, and other devices. (Wait, why does this sound familiar?) Of course, the groundwork would need to be laid out and discussed, which will probably cost us a few years...too bad we don't have a plan that could be put into action sooner, or maybe even was already deployed.
Anyway, the needless ROT13 text fairly well convinced me that our messages may be traveling over an ethernet bridge.
Please read and explain *exactly* why it doesn't work...
W
:
I'm wondering (and that shows that I have nothing better to do at 3:30am
on Monday...) how many people around here realize that the plain old
IPv4 - as widely implemented and specified in standard RFCs can be
easily used to connect pretty much arbitrary number (arbitrary means
2^256) of computers WITHOUT NETWORK ADDRESS TRANSLATION. Yes, you hear
me right.
And, no, it does not require any changes any in the global routing
infrastructure - as implemented now, and most OS kernels (those which
aren't broken-as-designed, grin) would do the trick just fine. None of
that dual-stack stupidity, and, of course, no chicken-and-egg problem if
the servers and gateways can be made to respect really old and
well-established standards.
DNS and most applications would need some (fairly trivial) updating,
though, to work properly with the extended addressing; and sysadmins
would need to do tweaks in their configs since some mythology-driven
"security" can get in the way. But they don't have to do that en mass
and all at once.
The most obvious solution to the non-problem of address space shortage
is the hardest to notice, ain't it?
well... not that it gained any traction atall, but given
the actual size/complexity of the global interconnect mesh,
we -could- ease the transition timing by many years with the
following administrative change. No tricks, no OS hacks,
no changes to software anywhere.. just a bit of renumbering...
recipie:
the usable IPv4 ranges
RFC 1918
Step one: Invert RFC 1918 to define the global Internets interconnection
mesh.
Step two: make all other usable IPv4 space "private".
Serves 2,000,000 million clients w/o changing to a new protocol family.
well... not that it gained any traction atall, but given
the actual size/complexity of the global interconnect mesh,
we -could- ease the transition timing by many years with the
following administrative change. No tricks, no OS hacks,
no changes to software anywhere.. just a bit of renumbering...
recipie:
the usable IPv4 ranges
RFC 1918
Step one: Invert RFC 1918 to define the global Internets
interconnection
mesh.
Step two: make all other usable IPv4 space "private".
Serves 2,000,000 million clients w/o changing to a new protocol
family.
Enjoy!
--bill
And I fully expect that to be done at some point or another. Country
takes the entire 32bit address space for itself. You want to serve that
country? Fine, apply for an allocation out of their /0 and route to it
over v6.
And I fully expect that to be done at some point or another. Country
takes the entire 32bit address space for itself. You want to serve
that
country? Fine, apply for an allocation out of their /0 and route to it
over v6.
What happens when countries are formed from secession? Does one half have to renumber?
3 bits at prepend. We only have 7 of those and Antarctica likely
doesn't need several billion addresses anyway. Got some leftover for
the United Federation of Planets. (or whatever other
semi-practical use that may be dreamed up)
You could do the same type of thing with E.164 country code ideas, but
that may be a bit stranger and drive the need for more RIRs along the way.