> Working Smarter, Not Harder.
>
> I love that phrase. To me, it sums up the whole concept of thinking
> "outside the nine dots" to come up with the improbable, the unexpected
> ways to eke more out of whatever we have. For example, consider the
> incredible improvement in modem speeds over the past 25 years,
> increasing the maximum data rate over the same dial-up phone line(!)
> from about 11 characters per second to 56K modems' theoretical maximum
> throughput of about 21,200 characters per second. Not a bad increase.
How the heck did anyone get any work done at 11 characters per second?
> But of course that isn't enough bandwidth to satisfy us (it never is),
> since we stubbornly keep coming up with new ways to consume the
> Internet bandwidth we have -- and more (graphics, sound, video,
> collaboration, and who know what's next.) So let's assume for the
> moment that whether you're using this week's fastest modems, or
> data-over-cable-TV, or xDSL, or whatever, that you still want a faster
> data rate from your Internet connection.
Yeah! YEAH! Faster!! FASTER!!!
> You could wait for fiber to your door. Or, suppose we could "work
> smarter," and get more data through the pipe you already have...
Want fiber. Want lotsa fiber. Connected straight to my cerebral
cortex. There's plenty of room there, I assure you little squickers...
> According to the June 2 DataQuest QuickTakes, two Israeli companies,
See, here we go again with the Israelis. These people are productive
TWENTY FOUR HOURS A DAY. I need to take a sabbatical there or
something. Can tenured grad students get sabbaticals?
> Flash Networks and Run, are doing just that, coming up with ways to get
> -far- more out of the TCP/IP protocol.
All I'd like at this point is the ability to *use* TCP on the pentium
cluster in our lab at Caltech. They no want work, they just want bang
on drums all day.
> So much more that DataQuest calls their preliminary results
> "...astounding...," achieving "...as much as five times the speed in
> downloads and transmissions over existing networks." And in Flash
> Network's case, this is done without any hardware assist!
Plus, it's a spiffy floorwax and a delicious dessert topping...
> How do they do it? According to Flash, it's a combination of "advanced
> flow-control algorithms and fast communications session establishment,
> ...highly efficient error handling, ...fault-tolerant downloads..."
> that pick up where they were interrupted, and easy to optimize
> performance.
Of course, why didn't I think of it?
> I haven't seen a lot of detail on these ideas yet, and so don't have
> any insight into what negatives might be associated with their
> implementation. However, I've been in the IT industry for awhile, and
> there is indeed an established history of "getting some things for
> nothing." (Which of course is just a catchy way of saying that people
> keep coming up with brilliant innovative ways to harness the
> ever-increasing processing power to better exploit the bottleneck de
> jure.) But you can begin to make up your own mind. Some details on one
> of these companies can be found at http://www.flash-networks.com .
Darn, my DNS server cannot resolve this. Guess I'll have to wait for my
own silver bullet.
----
adam@cs.caltech.edu
We have no voodoo.
-- Dan Zimmerman on why they can get it to work and we cannot