RE: Son of IKE: A proposal for moving forward

["Hallam-Baker, Phillip" <pbaker@verisign.com>]

Actually Moore's law is irrelevant here, or rather a corollary is more
relevant.

No matter how fast desktop machines become the number of 8 bit processors
based on Z-80 and 6502 cores in use increases at the same rate as processor
speed, number of circuits or any other factor you might want to consider.

The reason for this is that 8 bit processors get cheaper and cheaper and get
put into more and more stuff. At some point they will be put into light
bulbs and such, they are already in some rechargable batteries.

The real question is what point are you going to make the cut off. I think
that we can actually be fairly precise here, anything that needs IPSEC is
going to also need the ability to communicate via IP. That immediately
excludes processors with 64Kb memory spaces from rational consideration (yes
I do know about the Media lab HTTP server built on a single hydrogren atom
etc.).

For the majority of interesting cases adding communication to a device is
likely to mean adding some form of wireless. At this point my bet is that
Bluetooth is irrelevant and 802.11b is likely the only standard we need to
consider which gives us a reasonably high CPU power to work from.


		Phill


> -----Original Message-----
> From: Theodore Ts'o [mailto:tytso@mit.edu]
> Sent: Thursday, June 13, 2002 8:15 PM
> To: Dan Harkins
> Cc: Michael Thomas; uri@bell-labs.com; Stuart Jacobs;
> ipsec@lists.tislabs.com
> Subject: Re: Son of IKE: A proposal for moving forward
> 
> 
> On Thu, Jun 13, 2002 at 02:21:30PM -0700, Dan Harkins wrote:
> > 
> > The handwaving is not over Moore's Law, it's over some vague
> > "cheap wireless widget on the horizon". In fact, it's almost beyond
> > handwaving to outright fearmongering. Both of the protocols under
> > consideration are less heavy-weight than IKE and IKE has been
> > implemented on PDAs and cellphones. 
> > 
> 
> Michael,
> 
> If you could give us some specifics about what sorts of CPU/memory
> resources you think these "cheap wireless widgets" will have, that
> would be very helpful.  Given that StrongARM and PPC processors are
> regularly used in embedded devices, and very sophisticated systems
> have been implemented on embedded devices (including full Linux
> systems and full IKEv1 systems --- heck, IBM demonstrated Linux
> running on a wristwatch!  :-), without some quantifiable limitations 
> of these future wireless devices, it's very hard to move forward.
> 
> 						- Ted
>