I'm unconcerned because the problems affecting satellites are isomophic to
those facing gigabit transcontinental lines: the 'fat pipe problem'. No
matter how fast (or slow) your transmission speed, it's still at least ~30
ms light delay coast-to-coast rtt. While a modem is lucky to send a few
bytes in that window, the ack-delay is effectively zero packets. When a
terabit line sends hundreds of thousands of packets within that light-delay
window, the acks back up and explode.
So the solutions are tractable, too: streaming acks, selective nacks,
better modeling of the transport mediums' underlying error rates and
self-recovering error-resistant encodings. All of this MUST be solved on
the ground for fiber/SONET/etc, so it should port right over to space. The
worst case cost is a session-layer TCP gateway.
That's why Teledesic may stillbe right for not investing a dime in sw
research and letting the rest of the Net solve their problems in the next
36 months :-)
PS. Fooey on NASA for not predicting these problems when they turned on
their satellite 155mbps net -- the response-window ack-delay is obvious to
the most casual observer....
RK
----------
> From: CobraBoy <tbyars@earthlink.net>
> To: FoRK@xent.w3.org
> Subject: NASA finds Internet transmission faulty in space
> Date: Saturday, October 05, 1996 1:20 AM
>
> Wednesday, October 2, 1996
>
>
> NASA finds Internet
>
>
> Many, especially those in developing countries, may
> never get linked.
>
> By Jube Shiver Jr.
>
> Los Angeles Times
>
> WASHINGTON - In a significant setback for the satellite industry, NASA
> scientists have discovered that a key transmission standard that is the
> foundation of communications over the Internet and corporate computer
> networks does not work well in space.
>
> The discovery could delay the satellite industry's ambitious efforts to
> offer high-speed Internet access to companies with remote plants or
> offices, as well as to Pacific Islanders and millions of others who can't
log
> on.
>
> During the next decade, the satellite industry plans to spend more than
> $20 billion to extend modern communications services such as telephone,
> video programing and computer networks to remote regions.
>
> If satellites cannot be made to work seamlessly with the Internet, tens
> of billions of dollars in additional investment in land-based
transmission
> lines would be required instead. Many developing countries may never
> get connected.
>
> "Allowing satellites to be a part of the Internet would provide huge
> benefits to companies and individuals - if we could make it work," said
> Daniel R. Glover, a National Aeronautics Space Administration project
> engineer in Cleveland who has been spearheading the investigation into
> the problem.
>
> The TCP, or Transmission Control Protocol, problem arises because of
> the vast distances the data must travel to reach satellites.
>
> Electronic data can take as long as a half a second to travel up and
down
> to Earth from geostationary satellites 25,700 miles up. Although that
kind
> of delay only causes a slight echo when placing a telephone call, for
> example, it can wreak havoc when two computers are trying to
> communicate.
>
> That's because the delay causes TCP to believe there is a backup in the
> network, which in turn disrupts the electronic acknowledgements the
> receiving computer sends to confirm that messages are being received
> correctly.
>
> In one NASA experiment, a super-fast technology that normally sends
> data at 155 megabits per second over fiber-optic lines was slowed to just
> 10 megabits per second when NASA used a satellite communications
> network.
>
> Teledisc, a Kirkland, Wash., company, is developing a $9 billion
> low-Earth-orbiting satellite system that it believes can overcome the
> problems.
>
> However, Shawn Ostermann, an assistant professor in the electrical
> engineering and computer science department of Ohio University, said his
> research found even satellites orbiting closer to Earth can be adversely
> affected.
>
> Internet operators say they will oppose any changes that solve the
> satellite problem but hurt performance over regular phone lines.
>
> "It would not be an acceptable solution to degrade the current
> performance of TCP in order to improve it for satellite use," said Fred
> Baker, a software executive at Cisco Systems Inc., who is chairman of
> the Internet Engineer Task Force, a standards-setting body. "We don't
> want to break it just so somebody in New Caledonia can have better
> Internet access."
>
> --
>
> Inertia makes morons of a lot of people.
> _______________________________________
> http://radioactive.net/BANDS/BLACKG/blackg.html
> If you want the party over, take the disc out of the stereo.
>
> <> tbyars@earthlink.net <>