System transit time, Latency -

[jdow]

On 2012/04/07 16:43, Bob Goodwin - Zuni, Virginia, USA wrote:
> On 04/07/2012 07:06 PM, Sam Sharpe wrote:
>> On 7 April 2012 23:41, Bob Goodwin - Zuni, Virginia, USA
>> <bobgoodwin at wildblue.net> wrote:
>>
>>> Putting that address into mtr system2 produces the same complaint.
>>>
>>> [bobg at box7 ~]$ mtr system2 --address 10.73.255.21
>>>
>>> Failed to resolve host: Name or service not known
>> Do you have a host called system2 on your network? That looks like mtr
>> can't look up the name "system2" which would be as expected if it
>> doesn't exist.
>>
>> I would expect you to just do "mtr cnn.com".
>>
>
> Yes, you are correct sir, I misunderstood the command, mtr
> cnn.com, etc. does work perfectly and again I am seeing the
> system delay of a bit more than 600 ms.
>

I'd be a little surprised if it was materially less than 600ms for a
double satellite hop to Clarke orbit and back. You are going four hops,
at least. The speed of light delay is 88,000 miles/186000 mps is just
about half a second. Then you have satellite processing delay (it's not
a so called bent pipe) and you have your modem delays. Finally you do
have some finite time for the actual packet transit at each post on the
path. It'll be small, but measurable.

When I was testing data mode on the Magnavox (now gone) MX3000 satcom
modem I made a crazy link - Torrance Ca to Goonhilly ground station to
Cambridge Mass looped from there back to Torrance via a TCP connection
all at 2400bps. (That's as fast as Inmarsat-M provided for at that time.)
By the time I got done looping it around latency had gotten to around 2.5
seconds. The fool link still worked using TCP over the SatCom link. The
Inmarsat people wanted me to test this. They were not sure it'd work and
I had accounts to make the test. It was an expensive and perversely
amusing hack.)

{^_^}