> -----Original Message-----
> From: Ryan Barrett
> Sent: Saturday, November 11, 2006 9:24 AM
> Subject: latency vs bandwidth (was RE: [p2p-hackers] Cross-
> platformdevelopment)
> 
> david patterson [1], a berkeley cs prof who "wrote the book" on computer
> architecture, did a study [2] on this recently and concluded that
> bandwidth has been increasing with the *square* of latency since 1980 (!).

Just to make sure I understand the implications of this, are you saying that
bandwidth *improvements* have been increasing with the square of latency
*improvements* since 1980?  (I assume this must be what you mean, as
bandwidth has gotten faster while latency has gone down.)

Also, do you mean that this standard rule applies to all data transfer
mediums (internet, LAN, wireless, CPU bus, etc) about equally?  A sort of
Moore's law for the relationship between latency and bandwidth improvements?

I guess it kinda makes sense because latency can only improve so much --
there's an asymptotic approach toward instant -- while bandwidth can improve
infinitely.

What would you say are the consequences of such a trend?  With TCP on my
mind I'd say this affects the core algorithms of the internet as even an
infinite bandwidth connection will take a long time to ramp up if the
latency is slow.

But in a broader sense, what does this mean for the future?  Greater use of
decentralized caching over central serving?  Or is this observation true but
not relevant?

It's an interesting point, just trying to figure out what it all means.

-david


_______________________________________________
p2p-hackers mailing list
[email protected]
http://lists.zooko.com/mailman/listinfo/p2p-hackers

Reply via email to