"stephen" <stephen_hope@xyzworld.com> wrote:
>> Okay, you would benefit! That's good Jeff. But global legal
>> file sharing would be improved to, so I would benefit also,
>> which *is* important... ;-)
>
>er - no.
>
>you only improve efficiency if both users are on the same satellite, and

Which could easily be the typical case. For example with any
given satellite covering Alaska, and that would be true in many
areas of the world.

I don't think anyone is suggesting that this technology is being
developed to provide links between major cities in the US.

>neither are the ground station (which is probably true for military users on
>the same service, but not for general home users).

Neither the military (with mobile combat teams) nor the average
potential user around the world are connected via high capacity
fiber to ground stations. Perhaps you live where that is true,
but the whole point of developing this type of technology is to
provide better service to users who do not have access via high
capacity fiber optics.

>much more likely general case is Internet, with 1 user via the satellite,
>and 1 either on a terrestrial network, or maybe another ISP and / or
>satellite.

That is the general case now. But it does not describe most of
the world, and bringing better Internet service to the rest of
the world is the intent.

>So if the ground station is reasonably close to the terrestrial network (ie
>1000's of Km, since the satellite paths are so long) - no real gain in
>efficiency at all for general Internet use.
>
>The other assumption here is that the speed of light delay causes most of
>the latency - channel contention and queuing seem to add up to a lot more in
>practice....

And that is in fact the case. The actually propagation delay is
less than 500 ms for a round trip. A typical minimum ping time
is about 550ms. The 100ms or so difference, minimum, is latency
in various routers and other transmission systems. For an end
user like myself, typical minimum ping times run just over 600ms
over a satellite link.

>So - corporate network? a lot of the big ones are data centre to branch
>networks - nearly all the data flows via a couple of sites with big server
>farms. No real gain there either.

That is because with existing technology it makes little sense
to design any other style of network. The latency between leaf
nodes is too high.

>and most corporate networks only use satellite where they have to. one i
>worked with had 17k sites - 350 on satellite where we didnt have any
>choice.....

Exactly; and of course in parts of the world the majority of all
telecommunications is satellite based.

>Finally - scale. Dont forget the bandwidth down from the satellite on 1 or
>several beams has to carry all downward packets - and a few high speed
>microwave links are not going to compare to the 10s of 10G fibre interfaces
>a real backbone router might connect to.

I don't see your point. First, for communications to my area it
makes no difference at all what capacity exists for fiber. We
have the town literally webbed with fiber, but there is no fiber
interconnectivity to anywhere else, and probably will not be for
decades to come. It is not an uncommon situation, and in fact
describes much of the world.

In regard to satellite downlinks, there is some very interesting
potential there, which I have no details about and no mention
has been made in this discussion. Is the router limited to
traffic within one transponder, or does it have individual
connectivity to each transponder. Regardless of what the
initial technology will be, eventually satellites will have
routers that interconnect each transponder. And the number of
transponders is likely to be increased too, possibly with
narrower bandwidth for each. Imagine a satellite with a
hundreds of small foot print transponders!

The possibilities get very interesting.

>> Well, there are problems. One is that NSA will have to buy a
>> whole transponder on each satellite just to handle the snooping.
>> Congress can fund that though... (now they steal bandwidth on
>> virtually every satellite, but they don't need to access the
>> internals).
>
>Why bother? If the satellite has the "keys" to the encrypted traffic, then
>the ground station can get them as well. Once you have the keys - well the
>downlinks are microwave transponders, so pointing accuracy is not exactly a
>laser spot size - more like a country...... just listen in with a big dish.

That of course requires many listening stations, with at least
one within the footprint of each transponder (which might be
mobile in future satellites). Much easier to simply require
that each satellite have a transponder dedicated to the spooks,
who can then connect, in the satellite, to any data stream they
wish.

--
Floyd L. Davidson <http://www.apaflo.com/floyd_davidson>
Ukpeagvik (Barrow, Alaska) floyd@apaflo.com