IPv4 address length technical design

Thu Oct 4 08:10:00 UTC 2012

Eugen Leitl wrote:

> Except that these will be pure photonic networks, and apart from optical
> delay lines for your packet buffer you'd better be able to make a routing
> (switching) decision

Seriously speaking, that is the likely future as 1T
Ethernet will be impractical.

The point is to use 1Tbps packet by encoding a packet
simultaneously over, for example, 100 wavelengths each
encoded at 10Gbps.

> while few bits of the header have streamed by your
> photonic router circuit.
> 
> There is no time for any table look-ups, obviously.

At 1Tbps, 500B packet is 4ns long, which is long enough
for full /24 (with limited support for /32) look up.

While wavelengths containing header information are used
for table look up and rewritten, rest of the wavelengths
are delayed.

What you can't use with fiber delay lines is hash table,
which means large number (beyond CAM capacity) of Ethernet
MAC addresses is unusable and IPv6 with current allocation
scheme is bad.

> And optical gates are *really* expensive, so better use few of
> them. And don't add too many gate delays, too.

That's one reason why we should use 1Tbps packets. A gate
should switch not 10Gbps but 1Tbps or faster data.

Another reason is that 1Tbps packet needs 100 times shorter
delay lines to buffer than 10Gbps ones.

> Above describes your setting for the next protocol. There is not
> a lot of leeway in design space, I'm afraid.

Just keep using IPv4.

						Masataka Ohta
PS

See ftp://chacha.hpcl.titech.ac.jp/IEEE-ST.ppt presented
at a summer topical of IEEE photonic society.