[afnog] The path to eliminating IPv4 in the backbone

[Andrew Alston]

Aahhh, and there is the trick.

The mass recovery of v4 space when you’re running thousands of MPLS devices – becomes of major interest.

Also for us – its about making sure the building blocks are in place – and getting feature parity is about applying pressure to the vendors – and what I’m finding in this space – vendors mmmm actually react and get things in play reasonably quickly if you stop buying certain devices because of lack of certain features.

Andrew


From: Mark Tinka [mailto:mark.tinka at seacom.mu]
Sent: 02 October 2017 15:35
To: Andrew Alston <Andrew.Alston at liquidtelecom.com>; afnog at afnog.org
Subject: Re: [afnog] The path to eliminating IPv4 in the backbone


On 2/Oct/17 11:24, Andrew Alston wrote:

I figured I’d drop a note that may be of interest to some of you.

Thanks, Andrew, for sharing your experiences.




Effectively though – it is now possible to remove V4 entirely from P routers without losing any functionality – providing you engineer it right.  So we’re making progress 😊

So as you've confirmed again, a number of MPLS-based applications still do not have native IPv6 support, while they do in IPv4. This issue is compounded when operators maintain not only equipment from different vendors, but also different equipment from the same vendor. As if that's not enough, different software with different levels of feature support from the same or different vendors, for the same or different equipment.

We, for example, are a triple-vendor network - Cisco, Juniper and Arista. Fair point, we only use Arista for Layer 2 core switching, so those boxes do not participate in our MPLS domain. However, we have different platforms from Cisco and Juniper in operation, each running their own versions of software that are at different levels of development re: MPLS applications over IPv6.

Unless one is bound to recover a significant amount of IPv4 space by converting one's backbone (core, edge, or both) to IPv6-only, I'd only consider doing MPLSv6 network-wide once there is parity within and between vendors. I'd probably spend more time delivering working IPv6 to customers, until that day.

I can imagine the cluster-mess enabling MPLS applications for IPv6-only that would ensue if anyone tried this with the current state-of-the-art.

Mark.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.afnog.org/pipermail/afnog/attachments/20171002/ae4f722b/attachment.html>