RFC 6975

Signaling Cryptographic Algorithm Understanding in DNS Security Extensions (DNSSEC)

📅 published July 2013, from ietf

Part 1 of 2 💖

RFC 6975 🎀

Signaling Cryptographic Algorithm Understanding in DNS Security Extensions (DNSSEC) ✨

Okay bestie, this is RFC 6975, a Standards Track doc from July 2013 (ISSN 2070-1721). It's from the Internet Engineering Task Force (IETF), written by S. Crocker of Shinkuro Inc. and S. Rose of NIST. The header doesn't list any RFCs it obsoletes or updates, so it's standing on its own 🧚

Abstract ✨

DNSSEC was built to give DNS data origin authentication and integrity protection using digital signatures 🔐 Those signatures can be made with different algorithms, and this doc specifies a way for validating end-system resolvers to tell a server which digital signature and hash algorithms they support. The whole point is letting new algorithm uptake show up in client code, so zone admins can tell when it's safe to finish an algorithm rollover in a DNSSEC-signed zone. Main character energy for rollovers 🦋

Status of This Memo 📜

This is an Internet Standards Track document ⭐

It's a product of the IETF and represents the consensus of the IETF community. It got public review and was approved for publication by the Internet Engineering Steering Group (IESG). For more on Internet Standards, check Section 2 of RFC 5741 💜

For the current status, any errata, and how to give feedback, go to http://www.rfc-editor.org/info/rfc6975 🌸

Copyright (c) 2013 IETF Trust and the persons identified as the document authors. All rights reserved. 🌈

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) as in effect on the date of publication. Please read these carefully, because they spell out your rights and restrictions for this document. Code Components pulled out of this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions, and they come without warranty as described in the Simplified BSD License 🫶

Table of Contents 🍓

Table of Contents

   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . . . 3
   2.  Requirements Language . . . . . . . . . . . . . . . . . . . . . 4
   3.  Signaling DNSSEC Algorithm Understood (DAU), DS Hash
       Understood (DHU), and NSEC3 Hash Understood (N3U) Using EDNS  . 4
   4.  Client Considerations . . . . . . . . . . . . . . . . . . . . . 5
     4.1.  Stub Resolvers  . . . . . . . . . . . . . . . . . . . . . . 5
       4.1.1.  Validating Stub Resolvers . . . . . . . . . . . . . . . 5
       4.1.2.  Non-validating Stub Resolvers . . . . . . . . . . . . . 6
     4.2.  Recursive Resolvers . . . . . . . . . . . . . . . . . . . . 6
       4.2.1.  Validating Recursive Resolvers  . . . . . . . . . . . . 6
       4.2.2.  Non-validating Recursive Resolvers  . . . . . . . . . . 6
   5.  Intermediate System Considerations  . . . . . . . . . . . . . . 6
   6.  Server Considerations . . . . . . . . . . . . . . . . . . . . . 7
   7.  Traffic Analysis Considerations . . . . . . . . . . . . . . . . 7
   8.  IANA Considerations . . . . . . . . . . . . . . . . . . . . . . 8
   9.  Security Considerations . . . . . . . . . . . . . . . . . . . . 8
   10. Normative References  . . . . . . . . . . . . . . . . . . . . . 8

1. Introduction 🌸

Okay so DNSSEC ([RFC4033], [RFC4034], and [RFC4035]) exists to give DNS data origin authentication and integrity protection with digital signatures. Each RRSIG Resource Record (RR) carries an algorithm code number that matches a DNSKEY RR. Those codes are how validators know which cryptographic algorithm made the signature 🔑

Same vibe for the Delegation Signer (DS) RRs and the Hashed Authenticated Denial of Existence (NSEC3) RRs. They put a hashed value in their resource record data (RDATA), and those hash algorithms also have code numbers, just like the signature algorithms. All three sets of algorithm codes (RRSIG/DNSKEY, DS, and NSEC3) live in their own unique IANA registries 📚

Here's the plan: validating end-system resolvers get a way to tell a server, inside a DNS query, which digital signature and/or hash algorithms they support. That happens through new Extension Mechanisms for DNS (EDNS0) options, described in Section 2 for use in the OPT meta-RR [RFC6891]. All three new EDNS0 option codes are OPTIONAL to implement and use 💖

These options are basically a survey 📊 Zone administrators can use them to see how well code supporting the newly deployed digital signature algorithm, DS hash, and NSEC3 hash algorithm used with DNSSEC is actually rolling out. An admin can decide to stop signing with a superseded algorithm once the server sees that a lot of its clients can accept the new one. Heads up though: this survey might run for years before a tipping point shows up 🙈

Also, this doc is NOT making a new process for adding algorithms to DNSSEC, and it doesn't decide which algorithms belong on any official mandatory or recommended list. It just gives a way for a client query to signal the set of algorithms and hashes it implements. That's the whole job 🧚

2. Requirements Language 💅

Here are the key words, and how they're meant to be read: The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [RFC2119] ✨

3. Signaling DNSSEC Algorithm Understood (DAU), DS Hash Understood (DHU), and NSEC3 Hash Understood (N3U) Using EDNS 🦋

The EDNS0 spec in [RFC6891] gives us a standardized way to add new options. Those options live in the RDATA of the OPT meta-RR. This doc defines three new EDNS0 options so a client can signal which digital signature and/or hash algorithms it supports. They work independently of each other, they MAY appear in any order in the OPT RR, and each option code can appear only once in an OPT RR 🍓

Here's how each option is laid out in the RDATA of the OPT RR from [RFC6891]:

       0                       8                      16
       +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
       |                  OPTION-CODE                  |
       +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
       |                  LIST-LENGTH                  |
       +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
       |       ALG-CODE        |        ...            /
       +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+

OPTION-CODE is the code for the signaling option, and there are three of them:

  • DNSSEC Algorithm Understood (DAU) option for DNSSEC digital signing algorithms. Its value is fixed at 5.
  • DS Hash Understood (DHU) option for DS RR hash algorithms. Its value is fixed at 6.
  • NSEC3 Hash Understood (N3U) option for NSEC3 hash algorithms. Its value is fixed at 7.

LIST-LENGTH is the length, in octets, of the list of digital signature or hash algorithm codes. Each algorithm code takes up a single octet 🧮

ALG-CODE is the list of assigned values for DNSSEC zone signing algorithms, DS hash algorithms, or NSEC3 hash algorithms (whichever one the OPTION-CODE says) that the client declares it supports. The order of the code values can be arbitrary, and they MUST NOT be used to infer preference. So don't read anything into the order, periodt 💅

If all three options show up in the OPT RR, the OPT RR could get pretty chunky in the DNS message. In practical terms though, all three are likely to take up 22-32 octets (an average of 6-10 digital signature algorithms, 3-5 DS hash algorithms, and 1-5 NSEC3 hash algorithms), counting the EDNS0 option codes and option lengths, in potential future examples 🌈

4. Client Considerations 🫶

A validating end-system resolver sets the DAU, DHU, and/or N3U option, or any combination of them, in the OPT meta-RR when it sends a query. It MUST also set the DNSSEC OK bit [RFC4035] so it can receive DNSSEC RRs in the response ✨

Heads up about PRIVATEDNS (253) and PRIVATEOID (254): those digital signature codes cover a potentially huge range of algorithms, so they probably aren't useful to a server. There's no good reason for a client to put them in its DAU list. Likewise, clients MUST NOT include RESERVED codes in any of the options. A client also doesn't have to list every algorithm it implements, and it MAY choose to only list the algorithms it wants to signal as understood 💜

Since the DAU, DHU, and/or N3U options only get set in the query, if a client sees them in the response, it doesn't need to do anything, and it MUST ignore the option values 🙈

4.1. Stub Resolvers 🍓

Stub resolvers usually lean on an upstream recursive server (or cache) for their responses. So the best way to set the DAU, DSU, and N3U options depends on whether the stub resolver decides to do its own validation 🧚

4.1.1. Validating Stub Resolvers 🔐

A validating stub resolver sets the DNSSEC OK (DO) bit [RFC4035] to say it wants extra DNSSEC RRs (i.e., RRSIG RRs) in the response. Validating stub resolvers SHOULD include the DAU, DHU, and/or the N3U option(s) in the OPT RR when they send a query 💖

4.1.2. Non-validating Stub Resolvers 🙅

The DAU, DHU, and N3U EDNS0 options MUST NOT be included by non-validating stub resolvers. Not on the guest list 🚫

4.2. Recursive Resolvers 💅

4.2.1. Validating Recursive Resolvers 👑

A validating recursive resolver sets the DAU, DHU, and/or N3U option(s) when it does recursion, based on its own algorithm lists and on any DAU, DHU, and/or N3U option lists in the stub client's query. When the recursive server gets a query with one or more of the options set, it MUST set the algorithm list for any outgoing iterative queries in that resolution chain to a union of the stub client's list and the validating recursive resolver's list. For example, if the recursive resolver's algorithm list for the DAU option is (3, 5, 7) and the stub's algorithm list is (7, 8), then the final DAU algorithm list would be (3, 5, 7, 8) 🧮

If the client included the DO and Checking Disabled (CD) bits but left out the DAU, DHU, and/or N3U option(s) in the query, the validating recursive resolver MAY include the option(s) with its own list in full. And if one or more of the options are missing, the validating recursive resolver MAY include the missing options with its own list in full 🌸

Validating recursive resolvers MUST NOT set the DAU, DHU, and/or N3U option(s) in the final response to the stub client. Those options stay on the query side only ✨

4.2.2. Non-validating Recursive Resolvers 🌙

Recursive resolvers that don't do validation MUST copy the DAU, DHU, and/or N3U option(s) they see in received queries, because those options show the wishes of the validating downstream resolver that sent the original query 💜

5. Intermediate System Considerations ✨

okay so intermediate proxies (see Section 4.4.2 of [RFC5625]) that understand DNS are RECOMMENDED to act like a comparable recursive resolver when they're handling the DAU, DHU, and N3U options. it's giving resolver energy 🧚

6. Server Considerations 🦋

when an authoritative server sees the DAU, DHU, and/or N3U option(s) in the OPT meta-RR in a request, it just follows its normal algorithm for servicing requests. no drama. the options MUST NOT trigger any special processing (e.g., RRSIG filtering in responses) on the server side. the server is NOT supposed to start acting differently just because it spotted these, bestie 💅

if the options are present but the DO bit is not set, the server doesn't do any DNSSEC processing. that includes recording the option(s) too 🙈

if the server sees one (or more) of the options set with RESERVED values, it MAY ignore recording those values. totally its call ✨

authoritative servers MUST NOT set the DAU, DHU, and/or N3U option(s) on any responses. these values only ever show up in queries, periodt 🔥

7. Traffic Analysis Considerations 🌸

so zone administrators who are planning, or already in the middle of, a cryptographic algorithm rollover should keep an eye on their DNS query traffic. they should record how many queries come in, whether the OPT RR is present in those queries, and what the values of the DAU/DHU/N3U option(s) are (if they're there at all). this is how you measure how much client code out there implements and signals specific algorithms. 👀

how to actually capture DNS traffic and measure new algorithm adoption is beyond the scope of this document, so that's a different conversation 🫶

zone admins who need to follow changes to their org's security policy (when it comes to which cryptographic algorithms they can use) can use this data to set milestone dates for doing an algorithm rollover. for example, they can check when older algorithms can be phased out without disrupting a significant number of clients. to keep the disruption to a minimum, admins should wait to finish an algorithm rollover until a large majority of clients signal that they recognize the new algorithm. this might take years, not months. patience is a whole aesthetic 🍓

also, clients that don't implement these options are likely to be older implementations, which probably wouldn't implement any newly deployed algorithm either. the vintage crowd 💜

8. IANA Considerations 🎀

the algorithm codes used to identify DNSSEC algorithms, DS RR hash algorithms, and NSEC3 hash algorithms have already been established by IANA. this document does not seek to alter that registry in any way. hands off, we love a stable registry 🙌

IANA has allocated option codes 5, 6, and 7 for the DAU, DHU, and N3U options, respectively, in the "DNS EDNS0 Option Codes (OPT)" registry. all three options have a status of "standard". official and everything ⭐

9. Security Considerations 💀

okay, this part is serious, so read closely 🥺

this document specifies a way for a client to signal its digital signature and hash algorithm knowledge to a cache or server. it is NOT meant to be a discussion on algorithm superiority, so nobody get any ideas 🙅‍♀️ the signals are optional codes contained in the OPT meta-RR used with EDNS. the goal of these options is to signal new algorithm uptake in client code, so zone administrators know when it's possible to complete an algorithm rollover in a DNSSEC-signed zone. that's the whole point 🔥

There is a possibility that an eavesdropper or server could infer the validator in use by a client by the presence of the AU options and/or algorithm code list. this information leakage in itself is not very useful to a potential attacker, BUT it could be used to identify the validator or narrow down the possible validator implementations in use by a client. if that validator has a known vulnerability, an attacker could exploit it. yikes 😭

10. Normative References 📚

these are the references the spec depends on, so keep them handy:

[RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
           Requirement Levels", BCP 14, RFC 2119, March 1997.

[RFC4033]  Arends, R., Austein, R., Larson, M., Massey, D., and S.
           Rose, "DNS Security Introduction and Requirements",
           RFC 4033, March 2005.

[RFC4034]  Arends, R., Austein, R., Larson, M., Massey, D., and S.
           Rose, "Resource Records for the DNS Security Extensions",
           RFC 4034, March 2005.

[RFC4035]  Arends, R., Austein, R., Larson, M., Massey, D., and S.
           Rose, "Protocol Modifications for the DNS Security
           Extensions", RFC 4035, March 2005.

[RFC5625]  Bellis, R., "DNS Proxy Implementation Guidelines",
           BCP 152, RFC 5625, August 2009.

[RFC6891]  Damas, J., Graff, M., and P. Vixie, "Extension Mechanisms
           for DNS (EDNS(0))", STD 75, RFC 6891, April 2013.

Authors' Addresses 🧚‍♀️

meet the authors, they're iconic 👑

Steve Crocker
Shinkuro Inc.
5110 Edgemoor Lane
Bethesda, MD  20814
USA

EMail: steve@shinkuro.com


Scott Rose
NIST
100 Bureau Dr.
Gaithersburg, MD  20899
USA

Phone: +1-301-975-8439
EMail: scottr.nist@gmail.com
📦 where this page comes from