The Next Steps Toward Increasing The Security of DNSSEC with Elliptic Curve Cryptography
How do we make DNSSEC even more secure through the use of elliptic curve cryptography? What are the advantages of algorithms based on elliptic curves? And what steps need to happen to make this a reality? What challenges lie in the way?
Over the past few months we’ve been discussing these questions within the community of people implementing DNSSEC, with an aim of increasing both the security and performance of DNSSEC. Ondřej Surý of CZ.NIC Labs has been leading the way both with writing Internet drafts (draft-ietf-curdle-dnskey-ed25519 and draft-ietf-curdle-dnskey-ed448) and also in helping to organize sessions at various events.
Here’s a brief view of where that discussion has and will be taking place:
- 9 March 2016 – a panel session at ICANN 55 DNSSEC Workshop in Marrakech, Morocco- (see below)
- 1 April 2016 – a panel session at DNS-OARC in Buenos Aires
- 5 April 2016 – a discussion of the drafts in the CURDLE Working Group at IETF 95
- 6/8 April 2016 – a discussion of another draft in the DNSOP Working Group to reduce usage of older DNSSEC crypto algorithms
- 23-27 May 2016 – a panel session at RIPE 72 in Copenhagen, Denmark
- 27 June 2016 – a proposed panel session at the ICANN 56 DNSSEC Workshop in Helsinki, Finland
Let me provide a quick overview of what happened at ICANN 55 and then explain a new Internet draft that came out of that experience.
ICANN 55 DNSSEC Workshop
At ICANN 55 in Marrakech, we had a panel that I moderated where we presented several different viewpoints about how we go about implementing new DNSSEC algorithms and what are the challenges. I started out with a presentation where I outlined some of the challenges in this set of slides:
I was then followed by four panelists (links are to the slide decks three of the four panelists had):
- Geoff Huston, APNIC — ECDSA P-256 support in DNSSEC-validating Resolvers
- Jim Galvin, Afilias
- Ólafur Guðmundsson, CloudFlare — Adding New DNSSEC Algorithms: Reality Check
- Ondřej Surý, CZNIC — New Curves in DNSSEC
Geoff started out giving an overview of what APNIC’s research had found in the support of a current elliptic curve algorithm (ECDSA) in DNS resolvers (remembering that there are two sides to DNSSEC). Jim Galvin then provided a view of DNSSEC algorithms from a registry perspective. Olafur reported on the experience CloudFlare had rolling out ECDSA support and Ondřej wrapped up the session explaining the two new elliptic curve algorithms proposed for DNSSEC. There were a good number of questions asked and it was a healthy discussion.
Our Internet Draft on new deploying DNSSEC algorithms
After that ICANN 55 session, I went back and wrote up a summary of what we learned out of that discussion and then incorporate further input from Ondřej, Ólafur and Paul Wouters and turned that into a new Internet-draft:
As I said in the abstract:
As new cryptographic algorithms are developed for use in DNSSEC signing and validation, this document captures the steps needed for new algorithms to be deployed and enter general usage. The intent is to ensure a common understanding of the typical deployment process and potentially identify opportunities for improvement of operations.
We are looking forward to further discussion – and welcome any and all feedback on the document.
The DNS-OARC panel on Friday, April 1
Which leads to a mention of the next discussion happening on this Friday, April 1, at the DNS-OARC 24th meeting happening in Buenos Aires right before IETF 95. The very last session from 1700-1745 ART (UTC-3) will be on “DNSSEC algorithm flexibility” . I’ll be moderating the panel again and the focus this time will be on software implementations and what needs to be done there to support more encryption algorithms. Ondřej will be part of the panel along with Paul Wouters (Red Hat), Evan Hunt (ISC / BIND) and several others.
I’m told their will be a live stream of the DNS-OARC session and it should be accessible from the DNS-OARC Google+ page. I’ll update this post once I have an exact URL.
Our goal with all of this work is to lay out a solid path forward to bringing strong elliptic curve algorithms to DNSSEC – and then making that plan a reality. The end goal is an even more secure DNSSEC infrastructure that brings about an even more trusted DNS.
We’d welcome your comments and assistance with this – please do send us comments on the Internet Draft (email addresses at the end) or comment here or on social media about any of this. We need many different people helping move this forward!
P.S. If you are not yet using DNSSEC, please visit our Start Here page to begin!