Jeff Liebermann <jeffl@comix.santa-cruz.ca.us> wrote:
>floyd@apaflo.com (Floyd L. Davidson) hath wroth:
>
>Hello again. It's been a long time since we've locked horns.

Locked horns? We're both too old to have anything left but
worn stubs where majestic antlers once grew...

>>However, satellite bandwidth is expensive, and this would
>>increase efficiency by nearly two times. Plus provide better
>>service (lower latency, which is significant for both voice and
>>video conferencing).
>
>Sure. You forgot all those wonderful peer to peer applications that I
>mentioned later. Personally, I think it's a great idea. I just don't
>see how it's going to be built and implimented. For example, the
>traffic going up and down is not convenient ethernet packets that a
>router can easily digest. It's a compressed, paralleled, and possibly
>encrypted 6MHz wide 256QAM(?) RF stream.
>
>In order for a Cisco router
>to decide where to send this or that packet, the satellite would need
>to demodulate the whole mess, extract the packets, inspect the headers
>(and contents), reassemble the extracted packets, re-compress, and
>re-modulate the RF signal, before sending it on its way. I have no

I don't know that all readers have the background to understand
your point, so I'm going to expand on it with some detail.

Current satellites are mostly of the "bent pipe" design. There
is a receive antenna which picks up a signal from the ground
station (6GHz for C band uplink). The signal is hetrodyned to
the downlink frequency (4GHz for C band downlink) and amplified
before being sent to the transmit antenna. That is it. It is a
very simple, relatively small, radio equipment package.

Jeff is pointing out that instead of the currently simple radio
and electronics, it is not just a router that must be added to
the satellite, but virtually the entire set of electronics that
now exists only at the ground station. The received signal must
be sent to a downconverter that puts each transponder into a
separate 70Mhz Intermediate Frequency channel, and then it has
to be demodulated/decoded to the various digital bit streams
(one $30,000 modem per bit stream in a ground station) to have
access to each individual packet in order to route it to a
different destination than the packet before or after it. This
is a seriously larger amount of equipment than just adding a
little old Cisco router! It would be several times as much
equipment as current systems use.

Using current technology something like that is going to be very
crude and have few capabilities. But in 20 years it will be
doing things we can barely imagine right now. Hence most of the
discussion about what it can do, is not what the first try will
produce, but what it will eventually lead to.

>doubt that it can be done, but are the alleged benifits worth the
>effort? For consumer satellite internet, probably not. For military,
>cost is no object.

I think that once the critical mass is reached, and the price is
low enough that even a few consumers can afford it, it will
instantly become a very significant service, the price will
drop, and it will become ubiquitous.

The entire US may be layered with fiber optic cable, but the
rest of the world is not.

>>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... ;-)
>
>Nice try. For what it's worth, I barely do any illegal file sharing.
>I'm too busy doing other illegal things.

That's a good point. Why bother with something that only
produces a stupid song or two. There's real money to be made
out there! (Especially for faithful Republicans; but maybe we
can change that when Hillary is the CinC.)

>I will confess to having
>downloaded some cracked commercial software so I could play with it,
>and one DVD image, which incidentally was garbage and took 2 days to
>download. I don't think I would benifit very much.
>
>Incidentally, if the benifits of the latency reduction were so great,
>what happened to all the "terrestial satellite" ideas, such as
>tethered aerostat balloons and airplanes flying donuts?

Poof... History.

>>>The problem is that the ISP never sees the peer-to-peer
>>>traffic as it never goes down to the ISP's ground stations. That
>>>means that the ISP can't easily do filtering, traffic management,
>>>sniffing, and abuse mitigation.
>>
>>Why not? It's all done in the router. Granted that it will be
>>more expensive because power on board a satellite is limited,
>>but I suspect they've got this down to some fairly low power
>>technology by now.
>
>Perhaps I have a limited imagination, but I'm visualizing all the
>features and functions of a Cisco router run by remote control over
>the same inband channel used by the subscribers. That means Netflow
>and SNMP as implimented with Open View or some other network
>management program. To avoid having this become the "bottleneck in
>the sky", constant monitoring and traffic shaping will probably be
>required. I just can't visualize how it's going to be done unless
>they also install the network management workstation inside the bird
>and run it by remote control.

I think most of the potential for this is predicated on having a
significant increase in the onboard power available. If they
have it, all sorts of things are possible. Without it, the
beast is going to be very clunky and primitive.

Of course reduced power consumption is equally significant.

>I agree on the fairly low power technology in routers. That's been a
>design goal for many years as data centers are doing battle with the
>electric bill and cooling. Seen any ISP quality Cisco routers that
>are battery operated?
>
>>Well, there are problems. One is that NSA will have to buy a
>>whole transponder on each satellite just to handle the snooping.
>
>Yep. Not just NSA, but whomever operates the system probably wants to
>do some snooping. With all the traffic going directly between users,
>and NOT going through the ground station, snooping is almost
>impossible. However, NSA has (my) money, so dedicating a management
>channel for selective sniffing, err.... quality monitoring, would not
>be impossible.

Hmmm... I suspect that is already a virtual legal requirement.

>Incidentally, my reading of the tests performed on the existing CLEO
>test system point to IPv6, yet another layer of complexity.
>
>>>Yep. More research, justification, politics, bribes, deals, awards,
>>>litigation, consortia, licensing, auctions, press releases and
>>>technology are necessary.
>
>>Yeah, that's *wrong* (because *my* friends aren't getting enough).
>
>My friends say they aren't getting enough, so you aren't going to get
>any.

See... those are *greedy* bastards. Mine friends are *needy* bastards.

(Isn't that what everyone who lobbies Congress says?)

I've spent my whole life wondering why I never seem to get to the
good stuff before the greedy bastards do...

>The problem is that this router in the sky would normally be a
>fairly quiet application of existing technology and deployed without
>fanfare. However, something different is happening this time, which
>seems to be turning it into press release fodder. I'm suspicious
>enough to suspect that there's something else going on here.

A really good point. And I've seen nothing that indicates why
they've made a big deal out of it.

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