Re: [gaia] Heterogeneity in network capacity: growing?

Michael Welzl <> Mon, 29 February 2016 19:58 UTC

Return-Path: <>
Received: from localhost ( []) by (Postfix) with ESMTP id E6D3A1B3AF8 for <>; Mon, 29 Feb 2016 11:58:21 -0800 (PST)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -1.905
X-Spam-Status: No, score=-1.905 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.006] autolearn=ham
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id iHolTrj1kFKT for <>; Mon, 29 Feb 2016 11:58:19 -0800 (PST)
Received: from ( [IPv6:2001:700:100:10::17]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by (Postfix) with ESMTPS id 989381B383C for <>; Mon, 29 Feb 2016 11:58:18 -0800 (PST)
Received: from ([]) by with esmtp (Exim 4.80.1) (envelope-from <>) id 1aaTxQ-0003N4-OA; Mon, 29 Feb 2016 20:58:16 +0100
Received: from ([] helo=[]) by with esmtpsa (TLSv1:DHE-RSA-AES256-SHA:256) user michawe (Exim 4.80) (envelope-from <>) id 1aaTxP-0003yY-HM; Mon, 29 Feb 2016 20:58:16 +0100
Content-Type: multipart/alternative; boundary="Apple-Mail=_242B352C-9094-41A4-9FD0-05213363DEBE"
Mime-Version: 1.0 (Mac OS X Mail 9.2 \(3112\))
From: Michael Welzl <>
In-Reply-To: <>
Date: Mon, 29 Feb 2016 20:58:13 +0100
Message-Id: <>
References: <> <> <> <> <> <>
To: Henning Schulzrinne <>
X-Mailer: Apple Mail (2.3112)
X-UiO-Ratelimit-Test: rcpts/h 6 msgs/h 2 sum rcpts/h 8 sum msgs/h 3 total rcpts 38812 max rcpts/h 54 ratelimit 0
X-UiO-Spam-info: not spam, SpamAssassin (score=-5.0, required=5.0, autolearn=disabled, HTML_MESSAGE=0.001, TVD_RCVD_IP=0.001, UIO_MAIL_IS_INTERNAL=-5, uiobl=NO, uiouri=NO)
X-UiO-Scanned: 818CB5E6BD7ED1940B8050F4F569A858CE20B3A8
X-UiO-SPAM-Test: remote_host: spam_score: -49 maxlevel 80 minaction 2 bait 0 mail/h: 2 total 419 max/h 14 blacklist 0 greylist 0 ratelimit 0
Archived-At: <>
Cc: "" <>, Manner Jukka <>
Subject: Re: [gaia] Heterogeneity in network capacity: growing?
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Global Access to the Internet for All <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Mon, 29 Feb 2016 19:58:22 -0000

Well … I do think this is interesting as it is when considering the evolution of TCP, which should *always* work - which is why I’m less interested in the congested case, as this is also (at least to some degree) a function of TCP’s behavior. I think “how many % of hosts still face bandwidths of X” is more important than “how much of a market share is this”, at least from my idealist academic ivory tower...


> On 29. feb. 2016, at 19.31, Henning Schulzrinne <> wrote:
> I'm not sure this is a well-defined question. For example, if you are running a video service, the prevalence of modem connections (still about 1% in the US - <>) is of somewhat limited interest except to the extent that they'd cut into your addressable market. Thus, the question might be better phrased as "how many people are you willing NOT to reach for a given application" in a particular market. Also, in many cases, high-speed wireless data (e.g., LTE) may be available, but data charges are sufficiently high that many people will be reluctant to use them for more than a few MB/month. 
> In many rural areas in the US (and elsewhere), a big problem is diurnal variation of Internet speed - you might get 3 Mb/s at 2 am, but get 20% packet loss at 5 pm.
> Thus, theoretically, the dynamic range keeps increasing since the lowest speeds never go away, but the top available speeds keep increasing, but that may be no more interesting than noting that the range of memory capacities now ranges from a few kB in a light switch to PB in a data warehouse.
> Henning
> On Mon, Feb 29, 2016 at 11:26 AM, Michael Welzl < <>> wrote:
> > On 29. feb. 2016, at 10.11, Manner Jukka < <>> wrote:
> >
> > Hi,
> >
> > I don’t have a published analysis to cite, have done similar comparisons many times in various presentations. I could run that distribution from the Netradar database for 2014 and 2015, just need to give me the X buckets where to allocate the samples.
> NetRadar data would definitely be nice as a data point!  About buckets, not sure I get it: I mean e.g. a CDF of bandwidth values that were seen overall in 2014 and 2015; that would show if there is a shift not only towards higher bandwidths but if the diversity als increases, I guess.
> If you need to have buckets to query, I guess it can be any reasonable low / high bandwidth values… the question is, do the low bandwidths disappear as quickly as the higher ones pop up?
> > Market analysis is a bit trickier, since measurements need to be mapped to a region somehow. IP address is one option if it can be trusted enough.
> *I* don’t actually care about markets much. I just wonder what range of bandwidths network protocols need to be able to operate over… is it always the same range or is the range growing (as I suspect)?
> Cheers,
> Michael
> _______________________________________________
> gaia mailing list
> <>
> <>