Re: [homenet] Let's make in-home ULA presence a MUST !?

Markus Stenberg <> Wed, 15 October 2014 08:02 UTC

Return-Path: <>
Received: from localhost ( []) by (Postfix) with ESMTP id B3F521A0410 for <>; Wed, 15 Oct 2014 01:02:42 -0700 (PDT)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: 1.387
X-Spam-Level: *
X-Spam-Status: No, score=1.387 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, PLING_QUERY=0.994, SPF_NEUTRAL=0.779, URIBL_RHS_DOB=1.514] autolearn=no
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id qc9IeCAzi-CU for <>; Wed, 15 Oct 2014 01:02:40 -0700 (PDT)
Received: from ( []) by (Postfix) with ESMTP id 8314F1A040B for <>; Wed, 15 Oct 2014 01:02:40 -0700 (PDT)
Received: from poro.lan ( by ( (authenticated as stenma-47) id 543C16C90021EF49; Wed, 15 Oct 2014 11:02:33 +0300
Content-Type: text/plain; charset=us-ascii
Mime-Version: 1.0 (Mac OS X Mail 7.3 \(1878.6\))
From: Markus Stenberg <>
In-Reply-To: <>
Date: Wed, 15 Oct 2014 11:02:26 +0300
Content-Transfer-Encoding: quoted-printable
Message-Id: <>
References: <> <> <> <20141014142746.GX31092@Space.Net> <> <20141014145930.GY31092@Space.Net> <> <> <> <>
To: Mikael Abrahamsson <>
X-Mailer: Apple Mail (2.1878.6)
Cc: Markus Stenberg <>, Ted Lemon <>, Pierre Pfister <>, Erik Kline <>, HOMENET Working Group <>, Gert Doering <>
Subject: Re: [homenet] Let's make in-home ULA presence a MUST !?
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Wed, 15 Oct 2014 08:02:43 -0000

On 15.10.2014, at 7.58, Mikael Abrahamsson <> wrote:
> On Tue, 14 Oct 2014, Ted Lemon wrote:
>> Right.  This is IPv4.  In IPv4 we typically use a NAT on the local wire, so we get the effect we are trying to achieve either by retaining stale GUAs or using ULAs on the local wire in homenets.  IPv4 also does not provide graceful renumbering, so if the ISP wants to force an IPv4 address change, they have no choice but to do flash renumbering.  The UX is not good.
> Just to be clear, I am against flash renumbering, I want to see renumbering done with 30-60 minute overlap at least. I however do see that we really really need to support renumbering. One way of making sure that support works is to expose applications to frequent renumbering.

So essentially IPv6 home network would be ultimately worse platform than IPv4 one?

Because no matter what ISP does, my IPv4 prefixes in my home _are_ stable. And IPv6 ones too (thanks to using statically configured tunnel, cough).

I would argue that flash renumbering is bad too, but I would also claim that overlap of less than session lifetime is not acceptable. .. some of my session lifetimes are in days (single TCP session). Or are we planning to ultimately deprecate TCP and advocate something else? MPTCP? Yes, I know typical home user has just HTTP requests, but I am not fond of idea of turning the network to short-lived-connections only model just because it seems currently the most popular way to do things.

Every time I hear about ISP-forced customer renumberings, the more I start to think that 1+ ULA prefixes per home is a MUST, not a SHOULD.