[NNagain] A Multifaceted Look at Starlink Performance

[NNagain] A Multifaceted Look at Starlink Performance

[the keyboard of geoff goodfellow]

[-- Attachment #1: Type: text/plain, Size: 2895 bytes --]

EXCERPT:

*A Multifaceted Look at Starlink Performance *

Nitinder Mohan∗ Technical University of Munich Germany
Andrew E. Ferguson∗ The University of Edinburgh United Kingdom
Hendrik Cech∗ Technical University of Munich Germany
Rohan Bose Technical University of Munich Germany
Prakita Rayyan Renatin Technical University of Munich Germany
Mahesh K. Marina The University of Edinburgh United Kingdom
Jörg Ott Technical University of Munich Germany

*ABSTRACT*
In recent years, Low-Earth Orbit (LEO) mega-constellations have ushered in
a new era for ubiquitous Internet access. The Starlink network from SpaceX
stands out as the only commercial LEO network with over 2M+ customers and
more than 4000 operational satellites. In this paper, we conduct a
first-of-its-kind extensive multi-faceted analysis of Starlink performance
leveraging several measurement sources. First, based on 19.2M crowdsourced
M-Lab speed tests from 34 countries since 2021, we analyze Starlink global
performance relative to terrestrial cellular networks. Second, we examine
Starlink’s ability to support real-time latency and bandwidth-critical
applications by analyzing the performance of (i) Zoom conferencing, and
(ii) Luna cloud gaming, comparing it to 5G and fiber. Third, we perform
measurements from Starlink-enabled RIPE Atlas probes to shed light on the
last-mile access and other factors affecting its performance. Finally, we
conduct controlled experiments from Starlink dishes in two countries and
analyze the impact of globally synchronized “15-second reconfiguration
intervals” of the satellite links that cause substantial latency and
throughput variations. Our unique analysis paints the most comprehensive
picture of Starlink’s global and last-mile performance to date.

*1 INTRODUCTION *
Over the past two decades, the Internet’s reach has grown rapidly, driven
by innovations and investments in wireless access [22, 46, 47] (both
cellular and WiFi) and fiber backhaul deployment that has interconnected
the globe [3, 8, 10, 24, 77]. Yet, the emergence of Low-Earth Orbit (LEO)
satellite networking, spearheaded by ventures like Starlink [65], OneWeb
[49], and Kuiper [4], is poised to revolutionize global connectivity. LEO
networks consist of megaconstellations with thousands of satellites
orbiting at 300–2000 km altitudes, promising ubiquitous low latency
coverage worldwide. Consequently, these networks are morphing into “global
ISPs” capable of challenging existing Internet monopolies [66], bridging
connectivity gaps in remote regions [36, 69], and providing support in
disaster-struck regions with impaired terrestrial infrastructure [21]...

[...]
https://arxiv.org/pdf/2310.09242.pdf
via
https://twitter.com/TMFAssociates/status/1762204942297952382

-- 
Geoff.Goodfellow@iconia.com
living as The Truth is True

[-- Attachment #2: Type: text/html, Size: 5438 bytes --]

[NNagain] starlink IXP peering progress

[Dave Taht]

One of the things I learned today was that starlink has published an
extensive guide as to how existing BGP AS holders can peer with them
to get better service. I am curious if there is a way to see how many
have peered already, how many they could actually peer with?, and
progress over time since inception.... is there a tool for that?

https://starlink-enterprise-guide.readme.io/docs/peering-with-starlink

This is increasingingly off topic and too technical for the nnagain
mailing list (aside from documenting its impact and importance for
network resilience), is there a better email list to discuss ixp
stuff?

(I will send a copy of this separately to nanog)


On Mon, Feb 26, 2024 at 3:32 PM the keyboard of geoff goodfellow via
Starlink <starlink@lists.bufferbloat.net> wrote:
>
> EXCERPT:
>
> A Multifaceted Look at Starlink Performance
>
> Nitinder Mohan∗ Technical University of Munich Germany
> Andrew E. Ferguson∗ The University of Edinburgh United Kingdom
> Hendrik Cech∗ Technical University of Munich Germany
> Rohan Bose Technical University of Munich Germany
> Prakita Rayyan Renatin Technical University of Munich Germany
> Mahesh K. Marina The University of Edinburgh United Kingdom
> Jörg Ott Technical University of Munich Germany
>
> ABSTRACT
> In recent years, Low-Earth Orbit (LEO) mega-constellations have ushered in a new era for ubiquitous Internet access. The Starlink network from SpaceX stands out as the only commercial LEO network with over 2M+ customers and more than 4000 operational satellites. In this paper, we conduct a first-of-its-kind extensive multi-faceted analysis of Starlink performance leveraging several measurement sources. First, based on 19.2M crowdsourced M-Lab speed tests from 34 countries since 2021, we analyze Starlink global performance relative to terrestrial cellular networks. Second, we examine Starlink’s ability to support real-time latency and bandwidth-critical applications by analyzing the performance of (i) Zoom conferencing, and (ii) Luna cloud gaming, comparing it to 5G and fiber. Third, we perform measurements from Starlink-enabled RIPE Atlas probes to shed light on the last-mile access and other factors affecting its performance. Finally, we conduct controlled experiments from Starlink dishes in two countries and analyze the impact of globally synchronized “15-second reconfiguration intervals” of the satellite links that cause substantial latency and throughput variations. Our unique analysis paints the most comprehensive picture of Starlink’s global and last-mile performance to date.
>
> 1 INTRODUCTION
> Over the past two decades, the Internet’s reach has grown rapidly, driven by innovations and investments in wireless access [22, 46, 47] (both cellular and WiFi) and fiber backhaul deployment that has interconnected the globe [3, 8, 10, 24, 77]. Yet, the emergence of Low-Earth Orbit (LEO) satellite networking, spearheaded by ventures like Starlink [65], OneWeb [49], and Kuiper [4], is poised to revolutionize global connectivity. LEO networks consist of megaconstellations with thousands of satellites orbiting at 300–2000 km altitudes, promising ubiquitous low latency coverage worldwide. Consequently, these networks are morphing into “global ISPs” capable of challenging existing Internet monopolies [66], bridging connectivity gaps in remote regions [36, 69], and providing support in disaster-struck regions with impaired terrestrial infrastructure [21]...
>
> [...]
> https://arxiv.org/pdf/2310.09242.pdf
> via
> https://twitter.com/TMFAssociates/status/1762204942297952382
>
> --
> Geoff.Goodfellow@iconia.com
> living as The Truth is True
>
> _______________________________________________
> Starlink mailing list
> Starlink@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/starlink



-- 
https://blog.cerowrt.org/post/2024_predictions/
Dave Täht CSO, LibreQos

Re: [NNagain] starlink IXP peering progress

[Fearghas Mckay]

[-- Attachment #1: Type: text/plain, Size: 1017 bytes --]

 On 27 Feb 2024 at 14:51:29, Dave Taht via Nnagain <
nnagain@lists.bufferbloat.net> wrote:

> One of the things I learned today was that starlink has published an
> extensive guide as to how existing BGP AS holders can peer with them
> to get better service. I am curious if there is a way to see how many
> have peered already, how many they could actually peer with?, and
> progress over time since inception.... is there a tool for that?
>
> https://starlink-enterprise-guide.readme.io/docs/peering-with-starlink



This is just standard BGP peering over route servers. It is not BGP over
your starlink satellite service, you need to have connectivity to an
appropriate IXP.

It tells you to use Peering dB to find the appropriate locations -
https://www.peeringdb.com/net/18747

So if you connect to one of the IXPs linked on that page you can peer with
Starlink, just like another BGP network that does public peering over
routes.

The IXPs may have local data on who has been peering over the route
servers.

f

[-- Attachment #2: Type: text/html, Size: 1790 bytes --]

Re: [NNagain] A Multifaceted Look at Starlink Performance

[rjmcmahon]

I'm curious at what's needed by iperf here to help with connection 
timeouts.

Bob
> EXCERPT:
> 
>> A Multifaceted Look at Starlink Performance
>> 
>> Nitinder Mohan∗ Technical University of Munich Germany
>> Andrew E. Ferguson∗ The University of Edinburgh United Kingdom
>> Hendrik Cech∗ Technical University of Munich Germany
>> Rohan Bose Technical University of Munich Germany
>> Prakita Rayyan Renatin Technical University of Munich Germany
>> Mahesh K. Marina The University of Edinburgh United Kingdom
>> Jörg Ott Technical University of Munich Germany
>> 
>> ABSTRACT
>> In recent years, Low-Earth Orbit (LEO) mega-constellations have
>> ushered in a new era for ubiquitous Internet access. The Starlink
>> network from SpaceX stands out as the only commercial LEO network
>> with over 2M+ customers and more than 4000 operational satellites.
>> In this paper, we conduct a first-of-its-kind extensive
>> multi-faceted analysis of Starlink performance leveraging several
>> measurement sources. First, based on 19.2M crowdsourced M-Lab speed
>> tests from 34 countries since 2021, we analyze Starlink global
>> performance relative to terrestrial cellular networks. Second, we
>> examine Starlink’s ability to support real-time latency and
>> bandwidth-critical applications by analyzing the performance of (i)
>> Zoom conferencing, and (ii) Luna cloud gaming, comparing it to 5G
>> and fiber. Third, we perform measurements from Starlink-enabled RIPE
>> Atlas probes to shed light on the last-mile access and other factors
>> affecting its performance. Finally, we conduct controlled
>> experiments from Starlink dishes in two countries and analyze the
>> impact of globally synchronized “15-second reconfiguration
>> intervals” of the satellite links that cause substantial latency
>> and throughput variations. Our unique analysis paints the most
>> comprehensive picture of Starlink’s global and last-mile
>> performance to date.
>> 
>> 1 INTRODUCTION
>> Over the past two decades, the Internet’s reach has grown rapidly,
>> driven by innovations and investments in wireless access [22, 46,
>> 47] (both cellular and WiFi) and fiber backhaul deployment that has
>> interconnected the globe [3, 8, 10, 24, 77]. Yet, the emergence of
>> Low-Earth Orbit (LEO) satellite networking, spearheaded by ventures
>> like Starlink [65], OneWeb [49], and Kuiper [4], is poised to
>> revolutionize global connectivity. LEO networks consist of
>> megaconstellations with thousands of satellites orbiting at
>> 300–2000 km altitudes, promising ubiquitous low latency coverage
>> worldwide. Consequently, these networks are morphing into “global
>> ISPs” capable of challenging existing Internet monopolies [66],
>> bridging connectivity gaps in remote regions [36, 69], and providing
>> support in disaster-struck regions with impaired terrestrial
>> infrastructure [21]...
>  [...]
> 
> https://arxiv.org/pdf/2310.09242.pdf
> 
> via
> https://twitter.com/TMFAssociates/status/1762204942297952382
> 
> --
> 
> Geoff.Goodfellow@iconia.com
> 
> living as The Truth is True
> _______________________________________________
> Nnagain mailing list
> Nnagain@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/nnagain

Re: [NNagain] starlink ixp peering progress

[Bill Woodcock]

[-- Attachment #1: Type: text/plain, Size: 3151 bytes --]



> On Feb 27, 2024, at 08:54, Dave Taht <dave.taht@gmail.com> wrote:
> One of the things I learned today was that starlink has published an extensive guide as to how existing BGP AS holders can peer with them to get better service.

Yes, essentially every AS does this.  The ones that follow best-practices tend to be pretty uniform:

https://pch.net/peering
https://aws.amazon.com/peering/policy/
https://www.bytedance.com/en/peering
https://peering.google.com/#/options/peering
https://openconnect.netflix.com/en/peering/
https://www.lumen.com/en-us/about/legal/peering-policy.html
http://peering.ovh.net/

Starlink’s peering policy is straight-forward and follows best practices.  Then there are ones that get a little further afield, some of which can get kinda unusual in their fine-print:

https://learn.microsoft.com/en-us/azure/internet-peering/policy
https://www.verizon.com/business/terms/peering/
https://www.zayo.com/wp-content/uploads/2022/07/Zayo-Global-IP-Interconnection-Policy-Final-1.pdf
https://wholesale.orange.com/international/en/peering-policy.html

> I am curious if there is a way to see how many have peered already?

Well, if Starlink operates a looking-glass, sure.  Or you can derive an idea, albeit an incomplete one, from public data.  If any national communications regulators are paying attention, they may have placed a regulatory requirement on Starlink to make this public data, though it might have to be dug out of regulatory compliance filings, and might not be up-to-date.

> how many they could actually peer with?

That you’d need to script, though many people have.  We have an internal tool that tells us that about our own network, and I suspect pretty much every network large enough to have a dedicated peering team does likewise.  If you were to write such a tool for nonspecific use, we have public datasets that would show you who potential peers were at each IXP, and what routes / how many addresses they were advertising at each IXP…  Obviously if you’re learning Deutche Telekom’s routes in Frankfurt and Munich, it matters somewhat less whether you also peer with them in Karlsruhe, assuming they’re advertising the same routes everywhere, though it’s still good.  If they’re doing regional announcement, you might need to peer with them in different location to “collect the whole set” of their routes.  That’s not so common now, though it was a fad for a while, maybe fifteen years ago.  I haven’t tried to quantify the degree of regional announcement lately… that’s a good small project for a student who wants to learn about routing and interconnection.

> And progress over time since inception.... is there a tool for that?

I think you’d have to throw together your own tools for that, or derive it from public data such as the routing archives that we, RIPE, and Route-Views maintain.

> Is there a better email list to discuss ixp stuff?

The two I know of are ixp-discuss@pch.net and ixp-discuss@itu.int.  Both are pretty quiet, though both have very helpful people on them.

                                -Bill


[-- Attachment #2: Message signed with OpenPGP --]
[-- Type: application/pgp-signature, Size: 833 bytes --]