RE: Security for various IETF services

<> Wed, 09 April 2014 08:59 UTC

Return-Path: <>
Received: from localhost ( []) by (Postfix) with ESMTP id C590E1A01AF for <>; Wed, 9 Apr 2014 01:59:37 -0700 (PDT)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -1.9
X-Spam-Status: No, score=-1.9 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_NONE=-0.0001, SPF_PASS=-0.001, UNPARSEABLE_RELAY=0.001] autolearn=ham
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id IT7bV_khkXxL for <>; Wed, 9 Apr 2014 01:59:33 -0700 (PDT)
Received: from ( []) by (Postfix) with ESMTP id 059FE1A01AB for <>; Wed, 9 Apr 2014 01:59:32 -0700 (PDT)
Received: from [] by id 8C/0C-19700-4FB05435; Wed, 09 Apr 2014 08:59:32 +0000
X-Originating-IP: []
X-StarScan-Version: 6.11.1; banners=-,-,-
X-VirusChecked: Checked
Received: (qmail 28633 invoked from network); 9 Apr 2014 08:59:31 -0000
Received: from (HELO ( by with AES128-SHA encrypted SMTP; 9 Apr 2014 08:59:31 -0000
Received: from ([]) by ([]) with mapi; Wed, 9 Apr 2014 09:59:30 +0100
From: <>
To: <>, <>
Date: Wed, 9 Apr 2014 09:54:42 +0100
Subject: RE: Security for various IETF services
Thread-Topic: Security for various IETF services
Thread-Index: Ac9TzylyoALiO/RaQ+m7yme8v5VM7gAAi91U
Message-ID: <>
References: <> <> <> <> <> <> <> <> <> <011301cf532a$b4cd02a0$> <>, <>
In-Reply-To: <>
Accept-Language: en-US, en-GB
Content-Language: en-GB
acceptlanguage: en-US, en-GB
Content-Type: text/plain; charset="Windows-1252"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: IETF-Discussion <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Wed, 09 Apr 2014 08:59:38 -0000


the problem here is that the perpass work (also spearheaded by Stephen) has redefined what 'best practice' is.

perpass gives the security contingent of the IETF no limit in mitigation of pervasive monitoring -
which means making everything as secure and private as possible, even when common sense,
threat analysis and conventional 'best practice' would say otherwise.

perpass makes the IETF special.

Lloyd Wood
From: ietf [] On Behalf Of Yoav Nir []
Sent: 09 April 2014 09:38
To: Dick Franks
Cc: IETF-Discussion
Subject: Re: Security for various IETF services

On Apr 9, 2014, at 3:02 AM, Dick Franks <<>> wrote:

On 8 April 2014 09:32, t.p. <<>> wrote:

The path that I have seen several Security ADs steer Working Groups down
is to start with a threat analysis before deciding what counter measures
are appropriate.

Several contributors have been saying exactly that for almost a week.

These suggestions have been answered by dismissive emails and a relentless bombardment of magic pixie dust.

The thing is, the IETF websites are not that special. There’s a bunch of information, most of it public. The sites accept payments by credit card thrice a year. There’s a wiki. There’s some mailing lists with archives and registration pages, There’s content that can be edited by privileged users, and there’s other content that can be uploaded by anyone and viewed by anyone. Pretty much any corporate website has these parts.

The point of having best practices is that we don’t need to think about a threat model for each website. It’s a set of practices that someone has called “best practices” because they protect against several common types of attack at a low enough cost, that many web sites can implement them without actually proving to management that each part is strictly necessary.

Sure, we can have a debate here about the security needs of each service provided and have some of the most knowledgeable experts debate whether the<> should have HTTP access, while<> should have only HTTPS, and tools. should have both and HSTS and HPKP to boot. But that is not eating our own dog food, because the next “store that sells soap making supplies on the ‘net” will not have access to such expertise. Our dogfood should be implementable by people who don’t read these mailing lists - the web site developer that said store hires. Hence the need for “best practices” shortcuts, which are very much a part of all branches of engineering.

That said, if there are security implications that are not common to regular websites, these should be considered and discussed. For example, if reading drafts and RFCs from the office might betray your company’s future plans, that is a good reason to allow encrypted access to RFCs. Or if reading certain articles on Wikipedia can get you in trouble in some countries, then encrypted access to those may be in order. And if having both encrypted and non-encrypted access results in encrypted access being red-flagged as subversive, that may be a good argument for encrypted access by default or even exclusively.

But before diving into such a discussion we need to have a good argument for what makes us so special.