Re: [RAM] A curious Internet service offering

David Meyer <> Thu, 03 January 2008 01:39 UTC

Return-path: <>
Received: from [] ( by with esmtp (Exim 4.43) id 1JAF3K-0008IT-1K; Wed, 02 Jan 2008 20:39:22 -0500
Received: from [] ( by with esmtp (Exim 4.43) id 1JAF3I-0008IO-2D for; Wed, 02 Jan 2008 20:39:20 -0500
Received: from ([]) by with esmtp (Exim 4.43) id 1JAF3F-0006Zs-MS for; Wed, 02 Jan 2008 20:39:20 -0500
Received: from (localhost []) by (8.14.2/8.14.2/Debian-2) with ESMTP id m031dCss028296; Wed, 2 Jan 2008 17:39:12 -0800
Received: (from dmm@localhost) by (8.14.2/8.14.2/Submit) id m031dB0i028295; Wed, 2 Jan 2008 17:39:11 -0800
X-Authentication-Warning: dmm set sender to using -f
Date: Wed, 2 Jan 2008 17:39:11 -0800
From: David Meyer <>
To: RJ Atkinson <>
Subject: Re: [RAM] A curious Internet service offering
Message-ID: <>
References: <>
MIME-Version: 1.0
In-Reply-To: <>
X-gpg-fingerprint: 2409 8B50 B389 A307 BA5C 2A16 3918 03D6 A099 D8A7
X-philosophy: "Electric light still struck like arrows // Fired but for the ones // Condemned to drift or else be kept from drifting" -- Bob Dylan, "Chimes of Freedom"
User-Agent: Mutt/1.5.17 (2007-11-01)
X-Spam-Score: 0.0 (/)
X-Scan-Signature: a2c12dacc0736f14d6b540e805505a86
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Routing and Addressing Mailing List <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
Content-Type: multipart/mixed; boundary="===============1322785100=="

On Wed, Jan 02, 2008 at 11:13:43AM -0500, RJ Atkinson wrote:
> (NB: This doesn't directly relate to IRTF RRG work, but it does
> relate to routing & addressing futures, so the IAB RAM list
> seems to be the right venue for this narrow observation and
> any followup discussion that might occur.)
> I recently became aware of a large residential broadband operator
> in North America that provides no global-scope IP addresses to
> its customers.  By default there are no global-scope IP addresses
> -- and none are available as an option at any price to residential
> broadband subscribers to this particular service.
> Instead, this operator deploys a combination/integrated home
> gateway at each customer site.  This gateway is managed exclusively
> by the network operator.  The only customer option (at time
> of installation) is whether wireless is enabled or not.  This
> gateway performs NAT/NAPT, has an 802.11 wireless service on the
> customer side with WEP and WPA (but NOT 802.11i or WPA2), and
> uses DHCP to distribute private (RFC-1918; specifically 192.168.x/24)
> IP addresses to whatever devices the customer has on offer.
> This CPE box also includes a 4-port Ethernet hub on the inside
> of the NAT/NAPT to connect to any wired networks in the house.
> Further, there are sundry additional packet/port filters inside
> this CPE box.
> The net result is that this particular operator isn't really
> providing a "dialtone IP" service.  Instead, it is more nearly
> a "only web and email access" service.  For example, there are
> widespread reports that online gaming (e.g. using XBOX) does
> not work with this service.  There are also complaints online
> about how various uncommonly used transport-layer ports seem
> to be blocked.  The most commonly used ports (DNS, HTTP, HTTPS,
> IMAP4, SMTP, POP3) appear to work through this CPE box.  Of
> course, VoIP is also blocked -- though this operator does offer
> POTS lines via a separate adapter located at the customer premise.
> It is unclear to me whether/how this CPE integrated/combination
> home gateway is addressed.  One could imagine the CPE box being
> inside 10.0/8 and individual customers being inside 192.168.x/24
> with NAT/NAPT in the CPE box and then again at some larger gateway
> between the local region of this service and the public again.
> I don't know for certain whether the CPE box is addressed by
> IP, whether it has a private IP address, or whether it has a
> global-scope IP address.
> The operator has no issues with IPv4 address availability.  This
> is simply how they chose to define their service offering.  They
> market it as "High-speed Internet".  They believe that customers
> actually prefer to have the operator provide this narrower service
> rather than a "dial-tone IP" service.
> If this becomes a widely used deployment model, and customers accept
> this, then there are at least two implications to consider:
>   1) IPv4 Address shortages might not be as big an issue as some think.
>   2) New services really are only deployable over HTTP/HTTPS.
>      Nearly any other new protocol, NAT/NAPT-friendly or not,
>      would likely not be usable by these end users.


> I find the whole thing quite curious and unexpected.  I am sure
> that other folks mileage likely will vary somehwat from my own.

	Interestingly, Vince Fuller has been predicting this
	outcome for years.

RAM mailing list