> "It's absolutely a serious idea," he said today. "If we were successful in
> putting our rovers on the moon, we would put a server on the moon,
> too--but it could take $150 million [to get there]."
>
> The moon is 240,250 miles from the earth, so there could be a slight delay
> in serving up information, however. "There would be about a three-second
> round-trip delay for the signal to go back and forth," Nemitz noted.
Attribution, dammit. I'd quite like to read this article.
> -----
>
> So in other words, it will be faster than going thru Mae West?
Er, no.
The earth-moon distance varies slightly, but let's use the given
figure as a baseline. [Alternatively, you may wish to vote for your
own favourite Earth-Moon distance, but
http://www.astro.ku.dk/astronomyonline/market/collaboration/soleclipse/solecl-3d.html
has now closed. Tough.]
There and back (time) = 2 * distance / distance light travels in a second
= 2.58 seconds (neglecting latitude of earthpoint,
apparent latitude of moon,
extra delay in atmosphere, etc.
etc.)
But that's only if the server's on the moon, and the moon isn't just a
relay station signals get bounced off or regenerated at (an idea taken
quite seriously up until 1960 or so).
Otherwise, we have to double the round-trip time to give the time
taken for communication with an Earth server, via the moon, for a
single send of a packet of data. And then it takes the same time again
to send back something acknowledging that the data has been received
successfully.
Multiply this delay by the number of there-and-back acknowledgements
needed to even setup up a TCP connection (the handshake), consider the
window size you'd need to keep the channel full rather than waiting
for acknowledgements to come in so you can clear space for more data
in transit and how your TCP implementation probably doesn't quite
scale that far although it really ought to just as soon as you go out
and buy more memory - after all, what is RFC 1323 *for*? It's for the
*solar system*, that's what it's *for* - and suddenly the moon begins
to look decidedly second-rate for web surfing and as the place you'd
least want to locate your proxy cache on or use to conduct
communications. Unlike, say, Your Favourite Satellite Constellation,
which is nearer and Therefore Has Less Delay.
Frankly, the increase in speed of light in a vacuum over fibre and
copper just doesn't make up for the shorter lengths of fibre and
copper - it's a (relatively) long way to the moon.
On the other hand, if everyone's proxy server _was_ on the moon,
general internet congestion would be a lot less since less than half
the surface of the planet would be able to see the moon and surf the
web at any given time, and the equipment needed to bounce signals off
the moon would be expensive and heavily legislated as an EMC nuisance.
(The simplest way to get rid of congestion is always to pull the plug
on as many users as possible. Roads too crowded? Ban cars - problem
solved! networks too congested? Ban modems!)
The moon's effect on surfing as I've described above is perfectly
understandable - after all, you all know about its effect on tides.
The moon is a robust and mature implementation of a pull technology
known to be Y2K immune and to survive crashes, too - I think the delay
disadvantages could well be outweighed by these considerations.
L.
has the moon been patented?
<L.Wood@surrey.ac.uk>PGP<http://www.sat-net.com/L.Wood/>+44-1483-300800x3641