RFC 3445

Limiting the Scope of the KEY Resource Record (RR)

📅 published December 2002, from ietf

RFC 3445: Limiting the Scope of the KEY Resource Record (RR) 🎀

Bestie, this one is from December 2002 and it's a Standards Track doc 💅 It's from the Network Working Group, written by D. Massey (USC/ISI) and S. Rose (NIST). The header says it updates RFC 2535, and there's no Obsoletes line, so it isn't replacing anything. ✨

Status of this Memo 📜

This doc specifies an Internet standards track protocol for the Internet community, and it's literally asking for discussion and suggestions for improvements 🫶 For the standardization state and status of this protocol, check the current edition of the "Internet Official Protocol Standards" (STD 1). Distribution of this memo is unlimited. 🦋

Copyright (C) The Internet Society (2002). All Rights Reserved. 💜

Abstract 🌸

Okay so this doc limits the Domain Name System (DNS) KEY Resource Record (RR) to only keys used by the Domain Name System Security Extensions (DNSSEC). 💖 The original KEY RR used sub-typing to store both DNSSEC keys and arbitrary application keys, and storing both kinds under the same record type is a mistake 💀 This document removes application keys from the KEY record by redefining the Protocol Octet field in the KEY RR Data. Because application keys are gone, all but one of the flags in the KEY record become unnecessary and get redefined ✨ Three existing application key sub-types are changed to reserved, but the format of the KEY record is not changed. This document updates RFC 2535. 👑

1. Introduction 🧚

So this doc is all about limiting the scope of the KEY Resource Record (RR) 🎀 The KEY RR was defined in [3] and used resource record sub-typing to hold arbitrary public keys like Email, IPSEC, DNSSEC, and TLS keys. This document eliminates the existing Email, IPSEC, and TLS sub-types and prohibits new sub-types from being introduced. DNSSEC becomes the only allowable sub-type for the KEY RR, so sub-typing is basically eliminated, and all but one of the KEY RR flags are gone too 😭

Section 2 explains why we're restricting the KEY record, and Section 3 defines the revised KEY RR. Sections 4 and 5 summarize the changes from RFC 2535 and talk about backwards compatibility. It's really important to note that this document restricts how the KEY RR is used and simplifies the flags, but it does not change the definition or use of DNSSEC keys 🫶

Quick rule: the key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC 2119 [1]. 🙈

2. Motivation for Restricting the KEY RR 💅

The KEY RR RDATA [3] is made up of Flags, a Protocol Octet, an Algorithm type, and a Public Key. The Protocol Octet tells you which KEY RR sub-type you're looking at. DNSSEC public keys are stored in the KEY RR with a Protocol Octet value of 3. Email, IPSEC, and TLS keys were also stored in the KEY RR, using Protocol Octet values of 1,2, and 4 (respectively). Protocol Octet values 5-254 were available for assignment by IANA, and values were requested (but not assigned) for applications like SSH 🌈

Sub-typing has some inherent limitations, bestie 🙃 A resolver can not specify the sub-type it wants in a DNS query, and most DNS operations only apply to resource records sets. For example, a resolver can not directly request the DNSSEC subtype KEY RRs. Instead, it has to request all the KEY RRs for a DNS name and then dig through the set to find the DNSSEC sub-type it needs. DNSSEC signatures also apply to the whole set of KEY RRs for the DNS name, no matter the sub-type 💀

With the KEY RR, those sub-type limits get even worse, because the sub-type is what tells DNSSEC keys apart from application keys. These two kinds of keys differ in virtually every respect, and Section 2.1 breaks those differences down more. Mashing these very different keys into one sub-typed record adds unnecessary complexity and makes implementation and deployment errors more likely. Limited experimental deployment has shown that application keys stored in KEY RRs are problematic 🚩

So this document fixes all of that by removing every application key from the KEY RR 🔥 The scope here is strictly the KEY RR. This document does not endorse or restrict storing application keys in other resource records that haven't been defined yet 🦋

2.1 Differences Between DNSSEC and Application Keys 🔍

DNSSEC keys are an essential part of the DNSSEC protocol, and both name servers and resolvers use them to do DNS tasks. A DNS zone key, which is used to sign and authenticate RR sets, is the most common example of a DNSSEC key. SIG(0) [4] and TKEY [3] also use DNSSEC keys 👑

Application keys, like Email keys, IPSEC keys, and TLS keys, are just another type of data. They have no special meaning to a name server or a resolver 🎀

Here's the summary of how DNSSEC keys and application keys are different, lowkey the whole vibe of this section 💖

  1. They serve different purposes.
  2. They are managed by different administrators.
  3. They are authenticated according to different rules.
  4. Nameservers use different rules when including them in responses.
  5. Resolvers process them in different ways.
  6. Faults/key compromises have different consequences.

Now let's get into the tea, one by one ☕

  1. A DNSSEC key exists to sign resource records associated with a DNS zone (or to generate DNS transaction signatures in the case of SIG(0)/TKEY). An application key's purpose, though, is specific to the application. Application keys, like PGP/email, IPSEC, TLS, and SSH keys, aren't a mandatory part of any zone, and their purpose and proper use is outside the scope of DNS 🧚
  2. DNSSEC keys are managed by DNS administrators, but application keys are managed by application administrators. The DNS zone administrator decides the key lifetime, handles any suspected key compromises, and manages any DNSSEC key changes. The application administrator does those same jobs for the application keys tied to that application. Like, a user usually manages her own PGP key, and a server manages its own TLS key 🫶 Application key management is outside the scope of DNS administration.
  3. DNSSEC zone keys are used to authenticate application keys, but application keys are not allowed to authenticate DNS zone keys, by definition. 🙅 A DNS zone key is either configured as a trusted key or authenticated by building a chain of trust in the DNS hierarchy. To be part of that chain, a DNS zone needs to exchange zone key information with its parent zone [3]. Application keys aren't configured as trusted keys in the DNS and are never part of any DNS chain of trust. The parent doesn't need application key data, and it doesn't need to be exchanged with the parent zone for secure DNS resolution to work. A resolver treats an application key RRset as authenticated DNS information if it has a valid signature from the local DNS zone keys. Applications could still impose extra security requirements before they accept the application key as authentic for use with that application 🔐
  4. It may be useful for nameservers to include DNS zone keys in the additional section of a response, but application keys are typically not useful unless someone specifically asked for them. For example, it could be useful to include the example.com zone key along with a response that contains the www.example.com A record and SIG record, because a secure resolver needs the example.com zone key to check the SIG and authenticate the www.example.com A record. Including the IPSEC, email, and TLS keys along with the A record is typically not useful. Also, if you put application keys in the KEY record, a resolver would need the IPSEC, email, TLS, and other keys associated with example.com just to authenticate the example.com zone key, since signatures only apply to the entire KEY RR set. Depending on how many protocols are involved, the KEY RR set could get unwieldy for resolvers, and for DNS administrators to manage 😵‍💫
  5. DNS zone keys need special handling by resolvers, but application keys get treated the same as any other type of DNS data. DNSSEC keys aren't much use to end applications, unless those applications plan to do their own DNS authentication. By definition, secure resolvers are not allowed to use application keys as part of the authentication process. Application keys have no unique meaning to resolvers and are only useful to the application that asked for them. Also, if sub-types are used to identify an application key, then either the interface to the resolver has to specify the sub-type, or the application has to be able to accept all KEY RRs and pick out the one it wants 🌸
  6. A fault or compromise of a DNS zone key can lead to invalid or forged DNS data, but a fault or compromise of an application key should have no impact on other DNS data. Adding or changing a DNS zone key by mistake can invalidate all of the DNS data in the zone and in all of its subzones. An attacker who uses a compromised key can forge data from the affected zone and from any of its sub-zones. A fault or compromise of an application key has implications for that application, but it should not have an impact on the DNS. Note that application key faults and key compromises can hit the entire DNS if the application key and the DNS zone keys are both stored in the KEY RR 😱

Okay, the big takeaway: DNSSEC keys and application keys differ in almost every way 💅 DNSSEC keys are an essential part of the DNS infrastructure, and they need special handling from DNS administrators and resolvers. Application keys are just another type of data, with no special meaning to DNS administrators or resolvers. These two very different types of data do not belong in the same resource record. 🌈✨

3. Definition of the KEY RR 🎀

okay so the KEY RR uses type 25 and it's the record that stores DNSSEC keys 🔑 its RDATA is made up of flags, a protocol octet, the algorithm number octet, and the public key itself. here's the layout:

   ---------------------------------------------------------------------


                        1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 3 3
    0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |              flags            |   protocol    |   algorithm   |
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
   |                                                               /
   /                        public key                             /
   /                                                               /
   +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+

                             KEY RR Format

   ---------------------------------------------------------------------

in the flags field, all bits except bit 7 are reserved and MUST be zero 🙈 if Bit 7 (Zone bit) is set to 1, then the KEY is a DNS Zone key 👑 if Bit 7 is set to 0, the KEY is not a zone key. SIG(0)/TKEY are examples of DNSSEC keys that are not zone keys 🧚

the protocol field MUST be set to 3 💜

the algorithm and public key fields are not changed ✨

4. Changes from RFC 2535 KEY RR 🔄

the KEY RDATA format is not changed. literally the same shape 💅

okay so ALL the flags except the zone key flag got eliminated 😭 here's the tea:

  • The A/C bits (bits 0 and 1) are eliminated. They MUST be set to 0 and MUST be ignored by the receiver.
  • The extended flags bit (bit 3) is eliminated. It MUST be set to 0 and MUST be ignored by the receiver.
  • The host/user bit (bit 6) is eliminated. It MUST be set to 0 and MUST be ignored by the receiver.
  • The zone bit (bit 7) remains unchanged. iconic, we keep it 🎀
  • The signatory field (bits 12-15) are eliminated by [5]. They MUST be set to 0 and MUST be ignored by the receiver.
  • Bits 2,4,5,8,9,10,11 remain unchanged. They are reserved, MUST be set to zero and MUST be ignored by the receiver.

assigning any future KEY RR Flag values requires a standards action, so no sneaking new ones in 🙅‍♀️

and the protocol octet situation is just as dramatic 🎭 all Protocol Octet values except DNSSEC (3) are eliminated:

  • Value 1 (Email) is renamed to RESERVED. 📧➡️🚫
  • Value 2 (IPSEC) is renamed to RESERVED.
  • Value 3 (DNSSEC) is unchanged. the one and only queen 👑
  • Value 4 (TLS) is renamed to RESERVED.
  • Value 5-254 remains unchanged (reserved).
  • Value 255 (ANY) is renamed to RESERVED.

the authoritative data for a zone MUST NOT include any KEY records with a protocol octet other than 3. the registry IANA keeps for protocol values is closed for new assignments 🔒

name servers and resolvers SHOULD accept KEY RR sets that contain KEY RRs with a value other than 3 🫶 but if out of date DNS zones have deprecated KEY RRs with a protocol octet value other than 3, then just dropping those deprecated KEY RRs from the KEY RR set would invalidate any associated SIG record(s) and could create caching consistency problems 😩 the vibe is: be careful, bestie.

and one more thing, very important 🚨 KEY RRs with a protocol octet value other than 3 MUST NOT be used to authenticate DNS data. never ever 💀

the algorithm and public key fields are not changed 🌸

5. Backward Compatibility 🦋

DNSSEC zone KEY RRs are not changed and they stay backwards compatible 🥰 a properly formatted RFC 2535 zone KEY would have all flag bits, other than the Zone Bit (Bit 7), set to 0 and would have the Protocol Octet set to 3. this remains true under the restricted KEY, so we're all good ✨

DNSSEC non-zone KEY RRs (SIG(0)/TKEY keys) are backwards compatible too, but the distinction between host and user keys (flag bit 6) is lost 😢 a little plot twist.

no backwards compatibility is provided for application keys 🙃 any Email, IPSEC, or TLS keys are now deprecated. storing application keys in the KEY RR created problems like keys at the apex and large RR sets, and some change in the definition and/or usage of the KEY RR would have been needed even if this approach weren't adopted. it was a whole mess lol.

overall, existing nameservers and resolvers will keep correctly processing KEY RRs with a sub-type of DNSSEC keys 💖

6. Storing Application Keys in the DNS 🗝️

the scope of this document is strictly limited to the KEY record 🎯 this document prohibits storing application keys in the KEY record, but it does not endorse or restrict storing application keys in other record types. other documents can describe how DNS handles application keys, so go look there 📚

7. IANA Considerations 🧾

RFC 2535 created an IANA registry for DNS KEY RR Protocol Octet values. values 1, 2, 3, 4, and 255 were assigned by RFC 2535, and values 5-254 were made available for assignment by IANA. this document makes two sets of changes to this registry, so let's break it down 💅

first, this document re-assigns DNS KEY RR Protocol Octet values 1, 2, 4, and 255 to "reserved". DNS Key RR Protocol Octet Value 3 remains unchanged as "DNSSEC" 🔐

second, new values are no longer available for assignment by IANA, and this document closes the IANA registry for DNS KEY RR Protocol Octet Values 🚪 assigning any future KEY RR Protocol Octet values requires a standards action.

8. Security Considerations 🛡️

this document eliminates potential security problems that could arise from coupling DNS zone keys and application keys. before this change, a correctly authenticated KEY set could include both application keys and DNSSEC keys 😳 this document restricts the KEY RR to DNS security usage only. it's an attempt to simplify the security model and make it less prone to user error 🫠

if one of the application keys is compromised, it could be used as a false zone key to create false DNS signatures (SIG records). resolvers that do not carefully check the KEY sub-type could believe these false signatures and incorrectly authenticate DNS data. this is the dramatic part, bestie. with this change, application keys cannot appear in an authenticated KEY set, and this vulnerability is eliminated 🚨💀

the format and correct usage of DNSSEC keys is not changed by this document, and no new security considerations are introduced 🌈 phew.

9. Normative References 📖

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

   [2]  Eastlake, D., "Domain Name System Security Extensions", RFC
        2535, March 1999.

   [3]  Eastlake, D., "Secret Key Establishment for DNS (TKEY RR)", RFC
        2930, September 2000.

   [4]  Eastlake, D., "DNS Request and Transaction Signatures
        (SIG(0)s)", RFC 2931, September 2000.

   [5]  Wellington, B., "Secure Domain Name System (DNS) Dynamic
        Update", RFC 3007, November 2000.

10. Authors' Addresses 🧑‍💻

   Dan Massey
   USC Information Sciences Institute
   3811 N. Fairfax Drive
   Arlington, VA  22203
   USA

   EMail: masseyd@isi.edu


   Scott Rose
   National Institute for Standards and Technology
   100 Bureau Drive
   Gaithersburg, MD  20899-3460
   USA

   EMail: scott.rose@nist.gov

Copyright (C) The Internet Society (2002). All Rights Reserved.

This document and translations of it may be copied and furnished to others, and derivative works that comment on or otherwise explain it or assist in its implementation may be prepared, copied, published and distributed, in whole or in part, without restriction of any kind, provided that the above copyright notice and this paragraph are included on all such copies and derivative works. However, this document itself may not be modified in any way, such as by removing the copyright notice or references to the Internet Society or other Internet organizations, except as needed for the purpose of developing Internet standards in which case the procedures for copyrights defined in the Internet Standards process must be followed, or as required to translate it into languages other than English.

The limited permissions granted above are perpetual and will not be revoked by the Internet Society or its successors or assigns.

This document and the information contained herein is provided on an "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.

Acknowledgement 🌷

Funding for the RFC Editor function is currently provided by the Internet Society. 🫶

📦 where this page comes from