From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from oxalide-smtp-out.extra.cea.fr (oxalide-smtp-out.extra.cea.fr [132.168.224.13]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by lists.bufferbloat.net (Postfix) with ESMTPS id 2D7C83B2A4 for ; Wed, 5 Jun 2024 07:08:44 -0400 (EDT) Received: from e-emp-a0.extra.cea.fr (e-emp-a0.extra.cea.fr [132.167.198.35]) by oxalide-sys.extra.cea.fr (8.14.7/8.14.7/CEAnet-Internet-out-4.0) with ESMTP id 455B8gfA038470 for ; Wed, 5 Jun 2024 13:08:42 +0200 Received: from pps.filterd (e-emp-a0.extra.cea.fr [127.0.0.1]) by e-emp-a0.extra.cea.fr (8.18.1.2/8.18.1.2) with ESMTP id 4558Cukl013123 for ; Wed, 5 Jun 2024 13:08:42 +0200 Received: from muguet1-smtp-out.intra.cea.fr (muguet1-smtp-out.intra.cea.fr [132.166.192.12]) by e-emp-a0.extra.cea.fr (PPS) with ESMTP id 3yggacr0rv-1 for ; Wed, 05 Jun 2024 13:08:42 +0200 (MEST) Received: from [10.8.32.70] (is156570.intra.cea.fr [10.8.32.70]) by muguet1-sys.intra.cea.fr (8.14.7/8.14.7/CEAnet-Internet-out-4.0) with ESMTP id 455B8gGC003415 for ; Wed, 5 Jun 2024 13:08:42 +0200 Content-Type: multipart/alternative; boundary="------------tF0LukHSGDuJiHXPq98Ry20S" Message-ID: <688560aa-bd7c-441d-8fde-da287b50f59c@gmail.com> Date: Wed, 5 Jun 2024 13:08:42 +0200 MIME-Version: 1.0 User-Agent: Mozilla Thunderbird To: starlink@lists.bufferbloat.net References: Content-Language: fr From: Alexandre Petrescu In-Reply-To: X-Proofpoint-GUID: ymX6PDW-ZiJPQ_oexFSR9MBiqRmgNXsM X-Proofpoint-ORIG-GUID: ymX6PDW-ZiJPQ_oexFSR9MBiqRmgNXsM Subject: Re: [Starlink] musk: 28ms median latency on starlink X-BeenThere: starlink@lists.bufferbloat.net X-Mailman-Version: 2.1.20 Precedence: list List-Id: "Starlink has bufferbloat. Bad." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 05 Jun 2024 11:08:44 -0000 This is a multi-part message in MIME format. --------------tF0LukHSGDuJiHXPq98Ry20S Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit Le 04/06/2024 à 18:33, Livingood, Jason via Starlink a écrit : > > First question is what was the test destination? From the CPE to a > ground station or to something at a peering point or on the internet? > I would assume either to ground station or 1 hop beyond (peering > point)… That way we can do apples-to-apples comparisons, so to speak. > > In any case, there is a lot of good 3^rd party data out there for > this. First is RIPE Atlas. I helped get a lot of probes deployed in > the US and there are now 78 probes connected globally (roughly 30 > would be sufficient to call it statistically significant): > https://atlas.ripe.net/probes/public?sort=-id&toggle=all&page_size=100&search=AS14593&page=1&status=1 > . > So if you know how to pull data from Atlas, it is there for you to > analyze… (see https://atlas.ripe.net/docs/apis/rest-api-manual/ and > https://atlas.ripe.net/docs/tools-and-code/latencymon.html > > There is also this recent report that used a hardware probe connected > over ethernet: > https://www.netforecast.com/wp-content/uploads/FixedWireless_LEO_CableComparisonReport_NFR5148-1.pdf > > Another approach similar to accessing RIPE Atlas data would be to get > the Cloudflare AIM data from M-Lab – see > https://www.measurementlab.net/blog/cloudflare-aimscoredata-announcement/. > There was also a recent 3^rd party report on that (see Figure 1): > https://www.netforecast.com/wp-content/uploads/NFR5150_NetForecast-ISP-Performance-Report-2024.pdf. > > I was having the same question of where are the test destinations when talking about 28ms latencies. For my part, when I read latency evaluations expressed in milliseconds of RTT (round-trip time) I kind of assume we talk about using the tool of the website speedtest.net.  It is a very approximate number but I think it is understood by many people. That speedtest.net website has some server participants.  The test indicates where is situated the particular server in use. Of course, ideally, spacex should tell precisely how they obtained that latency number (28ms), but I doubt they'll ever get into that kind of detail.  It's as if I asked them to tell what algorithm is behind xai, or so. Alex > Jason > > *From: *Starlink on behalf of > Dave Taht via Starlink > *Reply-To: *Dave Taht > *Date: *Sunday, June 2, 2024 at 13:13 > *To: *Dave Taht via Starlink > *Subject: *[Starlink] musk: 28ms median latency on starlink > > Via elon musk: > > Starlink just achieved a new internal median latency record of 28ms > yesterday! Great work by the engineering and operations teams. > > - https://twitter.com/elonmusk/status/1797282250574184587 > > > I of course, am very interested in y'all´s external measurements of > how well starlink is doing. For me, it is fantastic - 30Mbit uploads > nowadays, 0 > > latency on the upload (how?) > https://www.waveform.com/tools/bufferbloat?test-id=2a1d139b-87cb-4ba4-a829-e2167801cffe > > > I also keep hoping that the rest of the ISP industry is now paying > attention and deploying stuff like fq_codel and cake and libreqos, > but, ah well - I will settle for starlink blowing past a lot of dsl > and cable and finding ways to get their density up. > > Anyone going to the Starship launch on the 6th? > > > -- > > https://www.youtube.com/watch?v=BVFWSyMp3xg&t=1098s > > Waves Podcast > > Dave Täht CSO, LibreQos > > > _______________________________________________ > Starlink mailing list > Starlink@lists.bufferbloat.net > https://lists.bufferbloat.net/listinfo/starlink --------------tF0LukHSGDuJiHXPq98Ry20S Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: 8bit


Le 04/06/2024 à 18:33, Livingood, Jason via Starlink a écrit :

First question is what was the test destination? From the CPE to a ground station or to something at a peering point or on the internet? I would assume either to ground station or 1 hop beyond (peering point)… That way we can do apples-to-apples comparisons, so to speak.

 

In any case, there is a lot of good 3rd party data out there for this. First is RIPE Atlas. I helped get a lot of probes deployed in the US and there are now 78 probes connected globally (roughly 30 would be sufficient to call it statistically significant): https://atlas.ripe.net/probes/public?sort=-id&toggle=all&page_size=100&search=AS14593&page=1&status=1. So if you know how to pull data from Atlas, it is there for you to analyze… (see https://atlas.ripe.net/docs/apis/rest-api-manual/ and https://atlas.ripe.net/docs/tools-and-code/latencymon.html

 

There is also this recent report that used a hardware probe connected over ethernet: https://www.netforecast.com/wp-content/uploads/FixedWireless_LEO_CableComparisonReport_NFR5148-1.pdf

 

Another approach similar to accessing RIPE Atlas data would be to get the Cloudflare AIM data from M-Lab – see https://www.measurementlab.net/blog/cloudflare-aimscoredata-announcement/. There was also a recent 3rd party report on that (see Figure 1): https://www.netforecast.com/wp-content/uploads/NFR5150_NetForecast-ISP-Performance-Report-2024.pdf.

I was having the same question of where are the test destinations when talking about 28ms latencies.

For my part, when I read latency evaluations expressed in milliseconds of RTT (round-trip time) I kind of assume we talk about using the tool of the website speedtest.net.  It is a very approximate number but I think it is understood by many people.  That speedtest.net website has some server participants.  The test indicates where is situated the particular server in use.

Of course, ideally, spacex should tell precisely how they obtained that latency number (28ms), but I doubt they'll ever get into that kind of detail.  It's as if I asked them to tell what algorithm is behind xai, or so.

Alex

 

Jason

 

 

From: Starlink <starlink-bounces@lists.bufferbloat.net> on behalf of Dave Taht via Starlink <starlink@lists.bufferbloat.net>
Reply-To: Dave Taht <dave.taht@gmail.com>
Date: Sunday, June 2, 2024 at 13:13
To: Dave Taht via Starlink <starlink@lists.bufferbloat.net>
Subject: [Starlink] musk: 28ms median latency on starlink

 

Via elon musk:

 

Starlink just achieved a new internal median latency record of 28ms yesterday! Great work by the engineering and operations teams.

 

- https://twitter.com/elonmusk/status/1797282250574184587

 

I of course, am very interested in y'all´s external measurements of how well starlink is doing. For me, it is fantastic - 30Mbit uploads nowadays, 0

 

I also keep hoping that the rest of the ISP industry is now paying attention and deploying stuff like fq_codel and cake and libreqos, but, ah well - I will settle for starlink blowing past a lot of dsl and cable and finding ways to get their density up.

 

Anyone going to the Starship launch on the 6th?

 


 

--

Dave Täht CSO, LibreQos


_______________________________________________
Starlink mailing list
Starlink@lists.bufferbloat.net
https://lists.bufferbloat.net/listinfo/starlink
--------------tF0LukHSGDuJiHXPq98Ry20S--