[IP] Not to beat a dead horse but a few more thoughts on bandwidth
Begin forwarded message:
From: "Jonathan M. Smith" <jms@xxxxxxxxxxxxxxxxxxxxx>
Date: January 3, 2006 7:08:37 PM EST
To: dave@xxxxxxxxxx
Cc: "Jonathan M. Smith" <jms@xxxxxxxxxxxxxxxxxxxxx>
Subject: Not to beat a dead horse but a few more thoughts on bandwidth
Dave:
I really enjoyed that discussion and it brought a few more
thoughts. First, an old rule of thumb in systems design (I may have
learned this from you): accountants care about throughput, users
care about response time. Second, something we thought about in
the Gigabit Testbed effort quite explicitly (e.g., in John Shaffer's
PhD thesis) was the relationship between these two things, which
is roughly that the response time (to get an object) is
propagation time + (object size) / throughput. Third, the
object sizes, due to both Moore's Law (memory sizes are typically
the fastest-rising exponential) and consumer demand (e.g., Mpixels
on digital cameras), are increasing rapidly. The speedups we achieved
with the testbed program achieved usability, as I mentioned in
a previous e-mail on this topic.
The takeaway from the response time relationship however,
is that an increase in throughput
commensurate with increases in memory and object sizes will be
necessary to maintain the present state of affairs, and a trajectory
of even greater throughput improvements will be necessary for
increased usability.
Best,
-JMS
------------------------------------------------------------------------
-
Jonathan M. Smith
Olga and Alberico Pompa Professor of Engineering and Applied Science
Professor of Computer and Information Science, University of
Pennsylvania
Levine Hall, 3330 Walnut Street, Philadelphia, PA 19104
-------------------------------------
You are subscribed as roessler@xxxxxxxxxxxxxxxxxx
To manage your subscription, go to
http://v2.listbox.com/member/?listname=ip
Archives at: http://www.interesting-people.org/archives/interesting-people/