[Starlink] 42 petabytes/day and ...

[Starlink] 42 petabytes/day and ...

[Dave Taht]

from here: https://news.ycombinator.com/item?id=39200323

There were two things that fell out of reading that article for me.

"each laser is grossly underused on average, at 0.432% of its maximum capacity."

+

"Brashears also said Starlink’s laser system was able to connect two
satellites over 5,400 kilometers (3,355 miles) apart. The link was so
long “it cut down through the atmosphere, all the way down to 30
kilometers above the surface of the Earth,” he said, before the
connection broke."

So there IS a way to achieve previously unheard of lower latencies (at
a cost in bitrate) across starlink across their network. Two hops to
go 10,000km.

I loved mark handley's original animation of how the ISL's were
supposed to work, but given the orbits  here, I kind of wish it was
easy to plug the assumptions in and figure out what the NY -> tokoyo
run would take in terms of hops and estimated switching overhead,
given this distance record.

How much data and what kind of data would benefit from that latency
reduction is a matter of speculation. "Buy! Sell!" between tokoyo and
london arbitrage was one of my first speculations many years ago.

-- 
40 years of net history, a couple songs:
https://www.youtube.com/watch?v=D9RGX6QFm5E
Dave Täht CSO, LibreQos

Re: [Starlink] 42 petabytes/day and ...

[Spencer Sevilla]

[-- Attachment #1: Type: text/plain, Size: 2196 bytes --]

Yeah I forget exactly where/when, but approx. five years ago there was a
LEO workshop at some big academic networking conference (maybe sigcomm?)
and I noticed that almost all the papers used NY-London latency as their
primary evaluation metric. One of the papers even proposed some wacky multi
hop system using commercial planes that were likely to be reliably
scheduled on the route. Confused the hell out of me (reading these papers
with an eye towards rural access) until my colleague pointed out the likely
funders of the research and their priorities 😂

On Thu, Feb 1, 2024, 22:07 Dave Taht via Starlink <
starlink@lists.bufferbloat.net> wrote:

> from here: https://news.ycombinator.com/item?id=39200323
>
> There were two things that fell out of reading that article for me.
>
> "each laser is grossly underused on average, at 0.432% of its maximum
> capacity."
>
> +
>
> "Brashears also said Starlink’s laser system was able to connect two
> satellites over 5,400 kilometers (3,355 miles) apart. The link was so
> long “it cut down through the atmosphere, all the way down to 30
> kilometers above the surface of the Earth,” he said, before the
> connection broke."
>
> So there IS a way to achieve previously unheard of lower latencies (at
> a cost in bitrate) across starlink across their network. Two hops to
> go 10,000km.
>
> I loved mark handley's original animation of how the ISL's were
> supposed to work, but given the orbits  here, I kind of wish it was
> easy to plug the assumptions in and figure out what the NY -> tokoyo
> run would take in terms of hops and estimated switching overhead,
> given this distance record.
>
> How much data and what kind of data would benefit from that latency
> reduction is a matter of speculation. "Buy! Sell!" between tokoyo and
> london arbitrage was one of my first speculations many years ago.
>
> --
> 40 years of net history, a couple songs:
> https://www.youtube.com/watch?v=D9RGX6QFm5E
> Dave Täht CSO, LibreQos
> _______________________________________________
> Starlink mailing list
> Starlink@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/starlink
>

[-- Attachment #2: Type: text/html, Size: 3013 bytes --]

Re: [Starlink] 42 petabytes/day and ...

[Mark Handley]

[-- Attachment #1: Type: text/plain, Size: 2982 bytes --]

In my original paper/video, I used NY-London as one of the key examples, not because of my funders (the work was actually unfunded - I just did it because I was curious), but because the only two applications I could immediately think of that cared enough about wide-area latency to pay for some premium service were finance and military.  I prefered not to write about the military uses, and finance routes like NY-Chicago are already covered by low-latency microwave towers.  Also I'm based in London :-)

Mark

On Fri, 2 Feb 2024, at 3:37 PM, Spencer Sevilla via Starlink wrote:
> Yeah I forget exactly where/when, but approx. five years ago there was a LEO workshop at some big academic networking conference (maybe sigcomm?) and I noticed that almost all the papers used NY-London latency as their primary evaluation metric. One of the papers even proposed some wacky multi hop system using commercial planes that were likely to be reliably scheduled on the route. Confused the hell out of me (reading these papers with an eye towards rural access) until my colleague pointed out the likely funders of the research and their priorities 😂
> 
> On Thu, Feb 1, 2024, 22:07 Dave Taht via Starlink <starlink@lists.bufferbloat.net> wrote:
>> from here: https://news.ycombinator.com/item?id=39200323
>> 
>> There were two things that fell out of reading that article for me.
>> 
>> "each laser is grossly underused on average, at 0.432% of its maximum capacity."
>> 
>> +
>> 
>> "Brashears also said Starlink’s laser system was able to connect two
>> satellites over 5,400 kilometers (3,355 miles) apart. The link was so
>> long “it cut down through the atmosphere, all the way down to 30
>> kilometers above the surface of the Earth,” he said, before the
>> connection broke."
>> 
>> So there IS a way to achieve previously unheard of lower latencies (at
>> a cost in bitrate) across starlink across their network. Two hops to
>> go 10,000km.
>> 
>> I loved mark handley's original animation of how the ISL's were
>> supposed to work, but given the orbits  here, I kind of wish it was
>> easy to plug the assumptions in and figure out what the NY -> tokoyo
>> run would take in terms of hops and estimated switching overhead,
>> given this distance record.
>> 
>> How much data and what kind of data would benefit from that latency
>> reduction is a matter of speculation. "Buy! Sell!" between tokoyo and
>> london arbitrage was one of my first speculations many years ago.
>> 
>> -- 
>> 40 years of net history, a couple songs:
>> https://www.youtube.com/watch?v=D9RGX6QFm5E
>> Dave Täht CSO, LibreQos
>> _______________________________________________
>> Starlink mailing list
>> Starlink@lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/starlink
> _______________________________________________
> Starlink mailing list
> Starlink@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/starlink
> 

[-- Attachment #2: Type: text/html, Size: 4668 bytes --]

Re: [Starlink] 42 petabytes/day and ...

[Alexandre Petrescu]

Le 02/02/2024 à 04:07, Dave Taht via Starlink a écrit :
> from here: https://news.ycombinator.com/item?id=39200323
>
> There were two things that fell out of reading that article for me.
>
> "each laser is grossly underused on average, at 0.432% of its maximum capacity."
>
> +
>
> "Brashears also said Starlink’s laser system was able to connect two
> satellites over 5,400 kilometers (3,355 miles) apart. The link was so
> long “it cut down through the atmosphere, all the way down to 30
> kilometers above the surface of the Earth,” he said, before the
> connection broke."

It looks spectacular.

A starlink latency of say 40ms on such a long distance 5400km between 
sats seems remarkable.  Note that the corresponding on-ground distance 
between two homes served by these two sats would be smaller, perhaps 
3000km(?).

It is easy to compare that via-stalink latency to a ping RTT on a ground 
only link on comparable distance.

Alex

>
> So there IS a way to achieve previously unheard of lower latencies (at
> a cost in bitrate) across starlink across their network. Two hops to
> go 10,000km.
>
> I loved mark handley's original animation of how the ISL's were
> supposed to work, but given the orbits  here, I kind of wish it was
> easy to plug the assumptions in and figure out what the NY -> tokoyo
> run would take in terms of hops and estimated switching overhead,
> given this distance record.
>
> How much data and what kind of data would benefit from that latency
> reduction is a matter of speculation. "Buy! Sell!" between tokoyo and
> london arbitrage was one of my first speculations many years ago.
>