Re: [tlug] [OT] Hope I get some on my next flight to Japan!

[Jim Breen <jimbreen@example.com>]

On Apr 4, 2014 1:15 PM, "Christian Horn" <chorn@example.com> wrote:
> On Thu, Apr 03, 2014 at 01:58:44PM -0300, [SCA] SCHWARTZ, Fernando G. wrote:
> > Dear Christian, indeed. Were they advertizing anything regarding
> > expected data throughput? I mean, any download/upload speed info?
>
> Not that, just
> - one story, multiple years old, crossing the atlantic ocean one would
>   see how the routes were changed in running "traceroute" in a loop

If we are talking about going via Inmarsat I would not expect many route
changes. AFAIK if you are using a particular Inmarsat service while
crossing the Atlantic, you'd be working with one geostationary the whole
time.

Of course speed will be low. It's the uplink that's the killer. (I could
dust off my lectures on link calculations if you like.) Plane to satellite
will be the lowest-capacity link, and in Internet-style networks the
delivered speed is that of the slowest hop. Always.

> - an other friend commented last year that the latency was horrible
>   (IIRC the RTT was 0.5 or 1sec), but he felt this was "just right" so
>   people would not try voice over IP, but one could work on mails.

Indeed. It *has* to be two satellite round trips. In the days when we
were still regularly putting voice over satellite links the rule was
one hop only or the latency made conversations impossible. A long
RTT doesn't really hurt file transfers; TCP should cope fine provided it
has big enough buffers and windows, and they all should by now.
The highly interactive apps, such as terminal emulators, heavily-
AJAXed WWW pages, etc. will range from awful to disastrous.

Cheers

Jim