[LINK] The PLAN, and broadband speeds?

[Richard Chirgwin]

Karl Auer wrote:
> Satellite delays are less noticeable for things like TV or web-pages,
> where a small outbound request results in a large download, but for any
> interactive application, satellite links are absolute poo-poo.
>   
Karl, I think there might be an excess of idealism in this statement.

Satellite links are acceptable for interactive apps. People can and do 
adjust for the delay.

TCP/IP is acceptable for interactive apps. People can and do adjust for 
its behaviour - it's not the physical layer that's the problem, but the 
latency inherent in routing that causes delay.

Satellite plus TCP/IP is a problem, but if it's a choice between 
satellite and nothing?

To put an idealist position - every place must have the best available 
and nothing less is acceptable - is simply not tenable. As for 
bandwidth, one-way satellite is no longer the only option. There are 
two-way birds up there now - IPstar, for example. Yes, you get delay - 
but it is at least a backfill for places that simply can't get any other 
connectivity.

It's also worth remarking that we don't have "copper everywhere" in 
Australia. We have copper "nearly everywhere". There are, according to 
the ACMA radio license database, several thousand remote Telstra 
customers whose "last X miles" connection is already a point-to-point 
VHF radio set. So it's probably worth remembering that such places exist 
even today...

RC
>   
>> Asymmetrical services can use terrestrial low-bandwidth back-channels
>>     
>
> Where you have these nasty latency problems.
>
> Regards, K.
>
>