[IPv6]Re: [v6ops] New draft: "The IPv6 Loopback Address Prefix"

["Templin (US), Fred L" <[e-mail]>]

Loopback (virtual) interfaces can be convenient handles for a node to park IPv6 GUAs that are not specific to any of its non-loopback interfaces. I would expect the use case discussion already present in the draft to say something about the common practice of assigning IPv6 GUAs to loopbacks and to clearly identify cases when a true loopback address from the prefix this draft proposes would be needed. For example, “when an IPv6 GUA is not available”, or “when communications need to be limited to the local node”, etc.

Also, the draft needs to update the RFC6742 Section 2.1 policy table to assign the correct precedence for the this new usage of ::/96 – the current precedence for this prefix is 1 and needs to be revised to 50.

Thank you - Fred

From: Warren Kumari <[e-mail]>
Sent: Wednesday, November 26, 2025 8:41 AM
To: Geoff Huston <[e-mail]>
Cc: IPv6 <[e-mail]>; IPv6 Operations <[e-mail]>
Subject: [IPv6]Re: [v6ops] New draft: "The IPv6 Loopback Address Prefix"

[ Replying mid-thread ]

Well, I was correct that there would be a fairly robust discussion on this topic, and it is still ongoing.

I am posting a new (-01) version with the /96 correction, but without any other (substantive) changes because there isn't anything that looks like clear consensus yet (AKA, we are not ignoring y'all's comments, we are just waiting for them them to coalesce…)

W


On Tue, Nov 25, 2025 at 2:20 PM, Geoff Huston <[e-mail]<mailto:[e-mail]>> wrote:
Hi Jeremy


So this one: 0:0:0:0:0:0:0:0 – 0:0:0:0:0:0:FFFF:FFFF

right?



You are of course correct - I was counting the number of bit to the right, where I should been counting the bits to the left

:-)

its a 96 - will be fixed in the next rev Thanks for spotting this!

Geoff






On 26 Nov 2025, at 4:55 am, Jeremy Duncan <[e-mail]<mailto:[e-mail]>> wrote:

Hi Warren/Geoff-

   ((Geoff: I have gone for a simple prefix that sits below the
   IPv4-mapped address block of 0:0:0:0:0:FFFF::/32 - the complication
   is that the prefix then includes the "unspecified address" as well.))

Regarding the IPv4-mapped address prefix, just to clarify, it’s a /96 not a /32 right?

The recommendation is saying that the loopback prefix will be a /96 not a /32 right?

So this one: 0:0:0:0:0:0:0:0 – 0:0:0:0:0:0:FFFF:FFFF

right?

Or are you saying 0:0:0:0:0:0:0:0 – 0:0:FFFF:FFFF:FFFF:FFFF:FFFF:FFFF

-Jeremy