[comp.protocols.tcp-ip] Variable length addresses

mo@prisma.UUCP (Mike O'Dell) (12/08/88)

Has anyone considered the performance hit by doing this?
My guess is that it will be non-trivial.  One could argue
that 96 bits is easily enough (fixed size) since that number
is big enough to enumerate some very sustantial fraction of all
the subatomic particles in the known universe.

	"If it's variable, it probably ain't fast."

		-Mike O'Dell

CERF@A.ISI.EDU (12/08/88)

Mike,

the performance question is what caused us to select a fixed
length, 32 bit address in the version 4 IP. Your question
is a good one and I wonder whether anyone has hard data for the
OSI IP version which does use a variable length address format.

It seems clear that we must give serious thought, though, to
a substantial increase in the number of networks addressable
within the IP scope if we are to scale to the size of the
global ISDN, for instance.

Vint

Makey@LOGICON.ARPA (Jeff Makey) (12/09/88)

In article <8812071620.AA06614@uunet.UU.NET> mo@prisma.UUCP (Mike O'Dell) writes:
>One could argue
>that 96 bits is easily enough (fixed size) since that number
>is big enough to enumerate some very sustantial fraction of all
>the subatomic particles in the known universe.

I whipped out the ol' calculator, and determined that 53 bits is more
than enough to enumerate each square foot of the Earth's surface.  How
many Earth-sized planets to we expect to populate in the future?  128
million?  Add another 27 bits for a total of 80.  Another 16 bits will
make room for 64K hosts per square foot (455 per square inch), and you
get a 96-bit internet address.

Not what I'd call a "very sustantial fraction of all the subatomic
particles in the known universe", but it sounds big enough for an
internet.  Just for safety, though, throw in an extra 32 bits for a
total of 128.  Nice, simple, fixed-length processing.  A mere 16
octets per address.  This should keep us out of trouble until at least
the year 2001.  :-)

                           :: Jeff Makey

Department of Tautological Pleonasms and Superfluous Redundancies Department
    Disclaimer: Logicon doesn't even know we're running news.
    Internet: Makey@LOGICON.ARPA    UUCP: {nosc,ucsd}!logicon.arpa!Makey