[AR] Re: Thinsats
- From: "JOHN HALPENNY" <dmarc-noreply@xxxxxxxxxxxxx> (Redacted sender "j.halpenny" for DMARC)
- To: "arocket@xxxxxxxxxxxxx" <arocket@xxxxxxxxxxxxx>
- Date: Mon, 26 Feb 2018 23:21:46 +0000 (UTC)
I thought one of the limits to battery life was the number of cycles. How long
do batteries last if they are cycled every 90 minutes?
John
On Monday, February 26, 2018 3:19 PM, Henry Spencer
<hspencer@xxxxxxxxxxxxx> wrote:
On Mon, 26 Feb 2018, Max Haot wrote:
In orbit, Gigabit connected, low latency orbit based storage, compute
cloud servers, live video could all become viable with small and
affordable modules for a given constellation...
I'm interested to find out where the flaw is in this thinking.
The question about such schemes is always "why isn't it better to put the
computing on the ground -- where mass, power, cooling, hardware upgrades,
maintenance, etc. are *much* cheaper -- and just use the satellites to
connect the users to the computing?".
The communications links are necessarily longer that way, but so long as
you avoid going via GSO, latencies can still be kept quite short. It's
unclear that there are any major virtues to be had by reducing latency
from (say) 10 milliseconds to 5.
A satellite constellation has one big inherent advantage: line of sight
to even remote users. So what it's really good for, is communication.
Whereas it stinks as a location for bulk computing resources. (The
question is not whether you can make it work somehow, but whether more
convenient locations would give you better service at lower cost. Sure
looks that way.)
For that matter, inexpensive lightweight solar-powered computing hardware
has uses on the ground too. No matter how low the latency to a satellite
is, the latency to the nearest cell tower or lamppost is a lot lower.
I'd be the last one to criticize LEO satellites, especially given that
much of my income comes from them now :-), but even if you make them
really cheap, there are applications they really truly aren't the best
choice for, and it sure looks to me like server farms are on that list.
Solar sat power will be the biggest limitation to the compute/transmission
power available but over time this is likely to improve with the added
benefit that it's free to use.
Alas, TANSTAAFW (there ain't no such thing as a free watt). :-) That
power is *not* free -- the solar cells to pick it up are costly to buy and
costly to put in position (and maintain there), and have a limited life.
(And all these issues except the first, double and redouble for the
batteries that store some of that solar power for the 40% of your orbit
that you spend in Earth's shadow.)
Henry
Other related posts: