| Internet Engineering Task Force | D. Wessels |
| Internet-Draft | P. Barber |
| Intended status: Standards Track | M. Weinberg |
| Expires: January 3, 2019 | Verisign |
| W. Kumari | |
| W. Hardaker | |
| USC/ISI | |
| July 2, 2018 |
Message Digest for DNS Zones
draft-wessels-dns-zone-digest-02
This document describes a protocol and DNS Resource Record used to provide a message digest over DNS zone data. In particular, it describes how to compute, sign, represent, and use the message digest to verify the contents of a zone for accuracy and completeness. The ZONEMD Resource Record type is introduced for conveying the message digest data.
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."
This Internet-Draft will expire on January 3, 2019.
Copyright (c) 2018 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 (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.
In the DNS, a zone is the collection of authoritative resource records (RRs) sharing a common origin ([RFC7719]), which can be distributed from primary to secondary name servers. Zones are often stored as files on disk in the so-called master file format [RFC1034]. Sometimes zones are distributed outside of the DNS, with such protocols as FTP, HTTP, rsync, and so on. Currently there is no standard way to verify the authenticity of a stand-alone zone file.
This document introduces a new RR type that serves as a cryptographic message digest of the data in a zone file. It allows a receiver of the zone file to verify the zone file's authenticity, especially when used in combination with DNSSEC. This technique makes the message digest a part of the zone file itself, allowing anything to verify the zone file as a whole, no matter how it is transmitted.
DNSSEC provides three strong security guarantees relevant to this protocol:
This specification is OPTIONAL to implement by both publishers and consumers of zone file data.
The motivation and design of this protocol enhancement is tied to the DNS root zone [InterNIC]. The root zone is perhaps the most widely distributed DNS zone on the Internet, served by 930 separate instances [RootServers] at the time of this writing. Additionally, many organizations configure their own name servers to serve the root zone locally. Reasons for doing so include privacy and reduced access time. [RFC7706] describes one, but not the only, way to do this. As the root zone spreads beyond its traditional deployment boundaries, the need for verification of the completeness of the zone contents becomes increasingly important.
One approach to preventing data tampering and corruption is to secure the distribution channel. The DNS has a number of features that can already be used for channel security. Perhaps the most widely used is DNS transaction signatures (TSIG [RFC2845]). TSIG uses shared secret keys and a message digest to protect individual query and response messages. It is generally used to authenticate and validate UPDATE [RFC2136], AXFR [RFC5936], and IXFR [RFC1995] messages.
DNS Request and Transaction Signatures (SIG(0) [RFC2931]) is another protocol extension designed to authenticate individual DNS transactions. Whereas SIG records were originally designed to cover specific RR types, SIG(0) is used to sign an entire DNS message. Unlike TSIG, SIG(0) uses public key cryptography rather than shared secrets.
The Transport Layer Security protocol suite is also designed to provide channel security. It is entirely possible, for example, to perform zone transfers using DNS-over-TLS ([RFC7858]). Furthermore, one can easily imagine the distribution of zone files over HTTPS-enabled web servers, as well as DNS-over-HTTPS [dns-over-https].
Unfortunately, the protections provided by these channel security techniques are ephemeral and are not retained after the data transfer is complete. They can ensure that the client receives the data from the expected server, and that the data sent by the server is not modified during transmission. However, they do not guarantee that the server transmits the data as originally published, and do not provide any methods to verify data that is read after transmission is complete. For example, a name server loading saved zone data upon restart cannot guarantee that the on-disk data has not been modified. For these reasons, it is preferable to secure the data itself.
DNSSEC provides certain data security guarantees. For zones that are signed, a recipient can validate all of the signed RRsets. Additionally, the denial-of-existence records can prove that RRsets have not been added or removed. However, not all RRsets in a zone are signed. The design of DNSSEC stipulates that delegations (non-apex NS records) are not signed, and neither are any glue records. Thus, changes to delegation and glue records cannot be detected by DNSSEC alone. Furthermore, zones that employ NSEC3 with opt-out are susceptible to the removal or addition of names between the signed nodes.
There are existing tools and protocols that provide data security, such as OpenPGP [RFC4880] and S/MIME [RFC3851]. In fact, the internic.net site publishes PGP signatures along side the root zone and other files available there. However, this is a detached signature with no strong association to the corresponding zone file other than its timestamp. Non-detached signatures are, of course, possible, but these necessarily change the format of the file being distributed. That is, a zone file signed with OpenPGP or S/MIME no longer looks like a zone file and could not directly be loaded into a name server. Once loaded the signature data is lost, so it does not survive further propagation.
It seems the desire for data security in DNS zones was envisioned as far back as 1997. [RFC2065] is an obsoleted specification of the first generation DNSSEC Security Extensions. It describes a zone transfer signature, aka AXFR SIG, which is similar to the technique proposed by this document. That is, it proposes ordering all RRsets in a zone, hashing their signatures, and then signing the zone hash. The AXFR SIG is described only for use during zone transfers. It did not postulate the need to validate zone data distributed outside of the DNS. Furthermore, its successor, [RFC2535], omits the AXFR SIG, while at the same time introducing an IXFR SIG.
This document introduces a new Resource Record type designed to convey a message digest of the content of a zone file. The digest is calculated at the time of zone publication. Ideally the zone is signed with DNSSEC to guarantee that any modifications of the digest can be detected.
The zone digest is designed to be used on zones that are relatively stable and have infrequent updates. As currently specified, the digest is re-calculated over the entire zone content each time. This specification does not provide an efficient mechanism for incremental updates of zone data. The authors believe that the incremental updates represent significant complexity which could be a barrier to implementation at this time (see DNS Camel). Nothing in this specification prevents future work to support incremental zone digest algorithms (e.g. using Merkle trees and a different RR type).
The cryptographic algorithms available for zone digest are exactly the same as for DS records. This avoids the need for a separate digest algorithm registry. Any updates to the DS algorithms automatically updates the algorithm status for zone digests.
It is expected that verification of a zone digest will be implemented in name server software. That is, a name server can verify the zone data it was given and refuse to serve a zone which fails verification. For signed zones, the name server would need a trust anchor for DNSSEC validation. For signed non-root zones, the name server may need to send queries to validate a chain-of-trust. Digest verification may also be performed externally.
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 BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.
This section describes the ZONEMD Resource Record, including its fields, wire format, and presentation format. The Type value for the ZONEMD RR is TBD. The ZONEMD RR is class independent. The RDATA of the resource record consists of three fields: Serial, Digest Type, and Digest.
FOR DISCUSSION: This document is currently written as though a zone MUST NOT contain more than one ZONEMD RR. Having exactly one ZONEMD record per zone simplifies this protocol and eliminates confusion around downgrade attacks, at the expense of algorithm agility.
The ZONEMD RDATA wire format is encoded as follows:
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
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Serial |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Digest Type | |
+-+-+-+-+-+-+-+-+ Digest +
/ /
/ /
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
The Serial field is a 32-bit unsigned integer in network order. It is equal to the serial number from the zone's SOA record ([RFC1035] section 3.3.13) for which the message digest was generated.
The Digest Type field is an 8-bit unsigned integer, with meaning equivalent to the Digest Type of the DS resource record, as defined in section 5.1.3 of [RFC4034].
The status of ZONEMD digest types (e.g., mandatory, optional, deprecated) SHALL always match the status for DS records. This information can be found in the IANA protocol registry for DS digest types [iana-ds-digest-types].
At the time of this writing the following digest types are defined:
| Value | Description | Status | Reference |
|---|---|---|---|
| 1 | SHA1 | Mandatory | [RFC3658] |
| 2 | SHA256 | Mandatory | [RFC4509] |
| 3 | GOST R 34.11-94 | Optional | [RFC5933] |
| 4 | SHA384 | Optional | [RFC6605] |
The Digest field is a variable-length sequence of octets containing the message digest. Section 3 describes how to calculate the digest for a zone. Section 4 describes how to use the digest to verify the contents of a zone.
The presentation format of the RDATA portion is as follows:
The Serial field MUST be represented as an unsigned decimal integer.
The Digest Type field MUST be represented as an unsigned decimal integer.
The Digest MUST be represented as a sequence of case-insensitive hexadecimal digits. Whitespace is allowed within the hexadecimal text.
The following example shows a ZONEMD RR.
example.com. 86400 IN ZONEMD ( 2018031500 4 FEBE3D4CE2EC2FFA4BA9
9D46CD69D6D29711E552
17057BEE7EB1A7B641A4
7BA7FED2DD5B97AE499F
AFA4F22C6BD647DE )
Calculation of the zone digest REQUIRES the RRs in a zone to be in a consistent format and ordering. Correct ordering of the zone depends on (1) ordering of owner names in the zone, (2) ordering of RRsets with the same owner name, and (3) ordering of RRs within an RRset.
This specification adopts DNSSEC's canonical ordering for names (Section 6.1 of [RFC4034]), and canonical ordering for RRs within an RRset (Section 6.3 of [RFC4034]). It also adopts DNSSEC's canonical RR form (Section 6.2 of [RFC4034]). However, since DNSSEC does not define a canonical ordering for RRsets having the same owner name, that ordering is defined here.
For the purposes of calculating the zone digest, RRsets having the same owner name MUST be numerically ordered by their numeric RR TYPE.
When AXFR is used to transfer zone data, the first and last records are always the SOA RR ([RFC5936] Section 2.2). Because of this, zone files on disk often contain two SOA RRs. When calculating the zone digest, the first SOA RR MUST be included and any subsequent SOA RRs MUST NOT be included.
Additionally, per established practices, the SOA record is generally the first record in a zone file. However, according to the requirement to sort RRsets with the same owner name by type, the SOA RR (type value 6) will not be first in the digest calculation. The zone's NS RRset (type value 2) at the apex MUST be processed before the SOA RR.
In preparation for calculating the zone digest, any existing ZONEMD records MUST first be deleted from the zone.
Prior to calculation of the digest, and prior to signing with DNSSEC, a placeholder ZONEMD record MUST be added to the zone. This serves two purposes: (1) it allows the digest to cover the Serial and Digest Type field values, and (2) ensures that appropriate denial-of-existence (NSEC, NSEC3) records are created if the zone is signed with DNSSEC.
In the placeholder record, the Serial field MUST be set to the current SOA Serial. The Digest Type field MUST be set to the value for the chosen digest algorithm. The Digest field MUST be set to all zeroes and of length appropriate for the chosen digest algorithm.
Following addition of the placeholder record, the zone MAY be signed with DNSSEC. Note that when the digest calculation is complete, and the ZONEMD record is updated, the signature(s) for that record MUST be recalculated and updated as well. Therefore, the signer is not required to calculate a signature over the placeholder record at this step in the process, but it is harmless to do so.
The zone digest is calculated by concatenating the canonical on-the-wire form of all RRs in the zone, in the order described above, subject to the inclusion/exclusion rules described below, and then applying the digest algorithm:
digest = digest_algorithm( RR(1) | RR(2) | RR(3) | ... )
where "|" denotes concatenation, and
RR(i) = owner | type | class | TTL | RDATA length | RDATA
When calculating the digest, the following inclusion/exclusion rules apply:
FOR DISCUSSION: Ambiguities about records that are in/out of zone. For example, see Jinmei message to dnsop 2018-06-01 and followups. BIND will load and axfr data "occluded" by DNAME/NS.
Once the zone digest has been calculated, its value is then copied to the Digest field of the ZONEMD record.
If the zone is signed with DNSSEC, the appropriate RRSIG records covering the ZONEMD record MUST then be added. Because the ZONEMD placeholder was added prior to signing, the zone will already have the appropriate denial-of-existence (NSEC, NSEC3) records.
Some implementations of incremental DNSSEC signing might update the zone's serial number for each resigning. However, to preserve the calculated digest, generation of the ZONEMD signature at this time MUST NOT also result in a change of the SOA serial number.
The recipient of a zone that has a message digest record can verify the zone by calculating the digest as follows:
This document uses a new DNS RR type, ZONEMD, whose value TBD has been allocated by IANA from the "Resource Record (RR) TYPEs" subregistry of the "Domain Name System (DNS) Parameters" registry.
The ZONEMD Digest Type field has the same semantics as the DS RR Digest Type field. Thus, it does not add new IANA protocol registry requirements.
The zone digest allows the receiver to verify that the zone contents haven't been modified since the zone was generated/published. Verification is strongest when the zone is also signed with DNSSEC. An attacker, whose goal is to modify zone content before it is used by the victim, may consider a number of different approaches.
The attacker might perform a downgrade attack to an unsigned zone. This is why Section 4 RECOMMENDS that the verifier determine whether or not to expect DNSSEC signatures for the zone in step 1.
The attacker might perform a downgrade attack by removing the ZONEMD record. This is why Section 4 REQUIRES that the verifier checks DNSSEC denial-of-existence proofs in step 2.
The attacker might alter the Digest Type or Digest fields of the ZONEMD record. Such modifications are detectable only with DNSSEC validation.
Nothing in this specification prevents clients from making, and servers from responding to, ZONEMD queries. One might consider how well ZONEMD responses could be used in a distributed denial-of-service amplification attack.
The ZONEMD RR is moderately sized, much like the DS RR. A single ZONEMD RR contributes approximately 40 to 65 octets to a DNS response, for currently defined digest types. Certainly other query types result in larger amplification effects (i.e., DNSKEY).
FOR DISCUSSION: The primary purpose of the ZONEMD record is to verify a zone file prior to being loaded or served by a name server. We could allow a name server implementation to respond to ZONEMD queries with the REFUSED RCODE without loss of functionality. Note that refusal would prevent ensuring that a zone-walk is complete.
This specification has no impacts on user privacy.
The authors wish to thank David Blacka, Scott Hollenbeck, and Rick Wilhelm for providing feedback on early drafts of this document. Additionally, they thank Mark Andrews, Olafur Gudmundsson, Shumon Huque, Tatuya Jinmei, Shane Kerr, Mukund Sivaraman, Petr Spacek, and other members of the dnsop working group for their input.
The authors are currently working on an implementation in C, using the ldns library [ldns]. This implementation is able to perform the following functions:
The authors expect to be able to release this implementation as open source following submission of this Internet-Draft.
RFC Editor: Please remove this section.
This section lists substantial changes to the document as it is being worked on.
From -00 to -01: