Starlink data starting to trickle in, despite NDAs on Beta Testers

Starlink data is starting to trickle in, despite NDAs on Beta Testers.

https://arstechnica.com/information-...acex-starlink-
beta-tests-show-speeds-up-to-60mbps-latency-as-low-as-31ms/

Bandwidth numbers from above:

Beta users of SpaceX's Starlink satellite-broadband service are
getting download speeds ranging from 11Mbps to 60Mbps, according to
tests conducted using Ookla's speedtest.net tool. Speed tests showed
upload speeds ranging from 5Mbps to 18Mbps.

Latency numbers from above:

The same tests, conducted over the past two weeks, showed latencies
or ping rates ranging from 31ms to 94ms.
....
Update at 11:18pm ET: A new Reddit post listing more speed tests
shows some Starlink users getting even lower latency of 21ms and
20ms.

Those are some decent numbers that will almost certainly blow rural DSL
providers out of the water.

Unfortunately, this Administration's FCC has been hostile towards
Starlink. From the article:

FCC Chairman Ajit Pai doubted Musk's latency claims and in May 2020
proposed to classify SpaceX and all other satellite operators as
high-latency providers?i.e. above 100ms?for purposes of a rural-
broadband funding distribution. The FCC backed off that plan but
said companies like SpaceX will have to prove they can offer low
latencies, and it continued to express "serious doubts" that
SpaceX and other similar providers will be able to deliver
latencies of less than 100ms.

Beta Testers are confirming that SpaceX's Starlink is *not* a "high
latency provider" like traditional geosynchronous satellite Internet
providers.

Jeff
--
All opinions posted by me on Usenet News are mine, and mine alone.
These posts do not reflect the opinions of my family, friends,
employer, or any organization that I am a member of.

In article ,
says...

On 2020-08-18 10:24, Jeff Findley wrote:

Beta users of SpaceX's Starlink satellite-broadband service are
getting download speeds ranging from 11Mbps to 60Mbps,

This is a meaningless number. What matters is how many customers will be
attached to a ground station and what is the capacity between ground
station and satellite.

Since when? It's pretty bog standard to run a speedtest like ookla. It
tests the bandwidth of your Internet connection, measured in Mbps over a
short span of time.

I do this when I suspect connection problems at home. Usually I get
more than 200 Mbps download and a few tens of Mbps upload. So, I would
gain nothing by using Starlink.

In a beta a test site likited to the seattle area with limited number of
beta testers, there is no way to know how close to max capacitty of the
one uplink (I assume to Starlink's offices) is used.

Which is why users will no doubt be running speed tests periodically to
see what they're getting. Obviously SpaceX already tracks all that
information.

Also, my cable Internet service is shared with everyone else in the area
that's also using that same service. Speeds can slow a bit during
periods of high usage. For example, like on a Friday night when
everyone on the planet seems to be using their streaming TV services.

Latency numbers from above:

Again, not relevant to real life. The current system is limited to use
only when you line within 1123km from the base station (the beta test is
stricter on distance). However, the original plan was relaxed to allow a
house to connect to a staellite that is 25° over horizon instead of 40°,
which greatly enlarged the footprint to a radius of 1123km below satellite.

Latency to servers users actually use (e.g. a gaming server) will get
better when the laser interlinks between satellites come online. When
that happens, it's most efficient to route packets through the
satellites and then down to the ground station closest to the
*destination*. That's *not* what's happening today.

The other issue will be where ground stations will be located. And
once
SpaceX launches satellites that hace inter saetllite links, that latensy
will need to be calculated, and whatecer jitter is introduced wheteber
your route in the air changes as satellites move around and the path
from you to ground station changes from one staellite to another and
between the satellite tyat gets your signal, and the staellites in the
path to A ground station.

That's still going to be a *lot* better than satellite Internet from a
geosynchronous satellite. I've known people that have that service.
It's absolutely awful in more ways than latency and bandwidth.

Jeff
--
All opinions posted by me on Usenet News are mine, and mine alone.
These posts do not reflect the opinions of my family, friends,
employer, or any organization that I am a member of.

On Wednesday, August 19, 2020 at 5:28:55 PM UTC-4, JF Mezei wrote:
On 2020-08-19 16:13, Jeff Findley wrote:

Since when? It's pretty bog standard to run a speedtest like ookla. It
tests the bandwidth of your Internet connection, measured in Mbps over a
short span of time.
A beta with a couple of users is not representative of what it will be
like in production because it does not represent the level of
oversubscription that Starlink intends to have to make the service
profitable.
Which is why users will no doubt be running speed tests periodically to
see what they're getting. Obviously SpaceX already tracks all that
information.
A fixed number of beta testers in a region with fixed number of base
station(1) will yield the same results and show stable service. They
already know how many periuods during the day there would be
interruptions in service and already know how each launch will reduce that.

But this beta allows Starlink to tests the residential base station
hardware and software that tracks satellites to orient antenna and
impact of westher. So Starlink is far more interested in signal strength
vs weather radar images than it would be of a speed test.

Note also that Starlink allowed base statiosn to connect to satellites
far lower above horizon in order to decreate the numebr of gaps in
service between satellite passes. So Starlink will be very interested in
knowing about signal strength at variious elevation of staellite vs
residential base station as this may help determine the eventual true
limit of how low above horizon a base station will attempt to look for a
staellite.
Also, my cable Internet service is shared with everyone else in the area
that's also using that same service.
And the cable company has extact statistics on node traffic for your
neighbourhood and when it reaches a certain level, begins process of
splitting the node into 2 nodes so that this happens before se5vice on
the 1 node degrades below its service standards.

Some cable companies have better standards than others.

Complaints from customers on slow speed tests do nothing since the
cableco has exact statistics on node usage taken at the CMTS port and
they know very well once customers start to see service degradation.
Latency to servers users actually use (e.g. a gaming server) will get
better when the laser interlinks between satellites come online.
Since the current fleet of satellites do not have the lasers physically
installed, it will be a long while before inter satellite links are of
use. And this will not really change latency. It is very doubtful that a
user in Port Henry NY connecting to a game server will have satellites
be aware that the destination is in Seattle and use intersatellite links
to bounde to a satellite that has view of the Seattle ground station
whyere it can had short ground hop to gaming server.

Such optmizations will likely be possible for permanent visrtual
circuits for enterprise customers (point to poing connections that
provide layer 2 connectivity) compared to normal Internet serfvice at
layer 3 which would require Satellites to have full Internet BGP routing
table and understand the concept of anycast to know where the nearest
servre is.
That's still going to be a *lot* better than satellite Internet from a
geosynchronous satellite.
These services are grossly oversubscribed in order to make them
profitable. And like with Starlink, they are limited in capacity of
ground ulink to statellite which is shared by all users. This is why
the numebr of ground stations that Starlink will deploy will be the
determining factor of how much capacity the system has. And if you are
in Port Henry and you are directed to a ground station in Albany, you
then travel via ground fibre to Seattle and will have worse service than
ground based Internet from your home which doesn't have the roughly
1000km hop up/down from satellite.

There is no question that LEO can theoretically beat the pants off a GEO
satellite service. But when you introduce the business aspect of how
much revenue you get and how much its costs, the question is still not
clear.

There are goo reasons why Starlink or any other satellite service
publish the uplink capacity for for their service. It is a key component
of how much oversubscription is needed to make service profitable and
how terrible the service will be.

Have you considered starting your own group "I Hate Elon Musk and Everything He is Affiliated With"?

On 8/18/2020 10:24 AM, Jeff Findley wrote:

Starlink data is starting to trickle in, despite NDAs on Beta Testers.

https://arstechnica.com/information-...acex-starlink-
beta-tests-show-speeds-up-to-60mbps-latency-as-low-as-31ms/

Bandwidth numbers from above:

Beta users of SpaceX's Starlink satellite-broadband service are
getting download speeds ranging from 11Mbps to 60Mbps, according to
tests conducted using Ookla's speedtest.net tool. Speed tests showed
upload speeds ranging from 5Mbps to 18Mbps.

Latency numbers from above:

The same tests, conducted over the past two weeks, showed latencies
or ping rates ranging from 31ms to 94ms.
...
Update at 11:18pm ET: A new Reddit post listing more speed tests
shows some Starlink users getting even lower latency of 21ms and
20ms.

Those are some decent numbers that will almost certainly blow rural DSL
providers out of the water.


Late to this thread but I can give some perspective vis my old
geosynchronous satellite service whom shall remain nameless but was the
biggest sat Internet provider of that time and is still a major player
today. Now this data is no doubt far behind what is available today. But
back in the year 2000 my satellite service was able to provide about
400kbps with a latency from 400 to 600ms. IIRC closer to 400 most of the
time. Then I learned a neat hack. The down-link servers in those days
would not limit the number of simultaneous connections you could have
across multiple ports. Thus using a Linux download program like wget and
specifying up to 5 simultaneous connections I was able to achieve (in
practice not theory) up to 1.5 Mbps of download capacity. Running in the
dead of night. Now in those days there was a hard download cap of 500MB.
Exceed that and your account got throttled back to 47kbps for a period
of 24 hours.

Just lending some perspective.

Dave

Oh yeah sorry I forgot uplink speed. Uplink was 28.8kbps all day long,
since it was provided by a dial-up provider....

Dave