| DNSOP | D. York |
| Internet-Draft | Internet Society |
| Intended status: Informational | O. Sury |
| Expires: May 4, 2017 | CZ.NIC |
| P. Wouters | |
| Red Hat | |
| O. Gudmundsson | |
| CloudFlare | |
| October 31, 2016 |
Observations on Deploying New DNSSEC Cryptographic Algorithms
draft-york-dnsop-deploying-dnssec-crypto-algs-02
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.
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 http://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 May 4, 2017.
Copyright (c) 2016 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) 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.
The DNS Security Extensions (DNSSEC), broadly defined in {{?RFC4033}}, {{?RFC4034}} and {{?RFC4035}}, make use of cryptographic algorithms in both the signing of DNS records and the validation of DNSSEC signatures by recursive resolvers.
The current list of cryptographic algorithms can be found in the IANA "Domain Name System Security (DNSSEC) Algorithm Numbers" registry located at <http://www.iana.org/assignments/dns-sec-alg-numbers/> Algorithms are added to this IANA registry through a process defined in {{?RFC6014}}. Note that {{?RFC6944}} provides some guidance as to which of these algorithms should be implemented and supported.
Historically DNSSEC signatures have primarily used cryptographic algorithms based on RSA keys. As deployment of DNSSEC has increased there has been interest in using newer and more secure algorithms, particularly those using elliptic curve cryptography.
The ECDSA algorithm {{?RFC6605}} has seen some adoption and two new algorithms are being proposed: Ed25519 and Ed448 .
The challenge is that the deployment of a new cryptographic algorithm for DNSSEC is not a simple process. DNSSEC algorithms are used throughout the DNS infrastructure for tasks such as:
In order for a new cryptographic algorithm to be fully deployed, all aspects of the DNS infrastructure that interact with DNSSEC must be updated to use the new algorithm.
This document outlines the current understanding of the components of the DNS infrastructure that need to be updated to deploy a new cryptographic algorithm.
It should be noted that DNSSEC is not alone in complexity of deployment. The IAB documented "Guidelines for Cryptographic Algorithm Agility" in {{?RFC7696}} to highlight the importance of this issue.
In this document, the key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" are to be interpreted as described in BCP 14, RFC 2119 .
For a new cryptographic algorithm to be deployed in DNSSEC, the following aspects of the DNS infrastructure must be updated:
Each of these aspects is discussed in more detail below.
DNS recursive resolvers perform "validation" to check the DNSSEC signatures of records received in a DNS query. To validate the signatures, the resolvers need to be able to understand the algorithm used to create the signatures.
In the case of a new algorithm, the resolver software needs to be updated. In some cases this could require waiting until an underlying library is updated to support the new algorithm.
Once the software is updated, the updates need to be deployed to all resolvers using that software. This can be challenging in cases of customer-premises equipment (CPE) that does not have any mechanism for automatic updating.
It should be noted that section 5.2 of {{?RFC4035}} states:
“If the resolver does not support any of the algorithms listed in an authenticated DS RRset, then the resolver will not be able to verify the authentication path to the child zone. In this case, the resolver SHOULD treat the child zone as if it were unsigned.”
This means that signing a zone with a new algorithm that is not widely supported by DNS resolvers would result in the signatures being ignored and the zone treated as unsigned until resolvers were updated to recognize the new algorithm.
Note that in at least one 2016 case (an ISP in Sweden) the resolver software deployed on customer premises turned out not to be compliant with RFC 4035. Instead of ignoring the signatures using unknown algorithms and treating the zones as unsigned, the validating resolver rejected the signatures and returned a SERVFAIL to the DNS query. This resulted in the ISP turning off DNSSEC validation on the equipment. Further investigation showed that a newer version of the resolver software did correctly support ECDSA, but now all customer premises equipment must be updated to this new version.
The point is that it is not safe to assume all resolver software will correctly implement this part of RFC 4035.
Authoritative DNS servers serve out signed DNS records. Serving new DNSSEC signing algorithms should not be a problem as a well-written authoritative DNS server implementation should be agnostic to the RR DATA they serve.
The one exception is if the new cryptographic algorithms are used in the creation of NSEC/NSEC3 responses. In the case of new NSEC/NSEC3 algorithms, the authoritative DNS server software would need to be updated to be able to use the new algorithms.
Note that some authoritative server implementations could include DNSSEC signing as part of the server and thus also fall into the "Signing Software" category below.
The software performing the signing of the records needs to be updated with the new cryptographic algorithm.
User interfaces that allow users to interact with the DNSSEC signing software may also need to be updated to reflect the existence of the new algorithm.
Note that the key and signatures with the new algorithm will need to co-exist with the existing key and signatures for some period of time. This will have an impact on the size of the DNS records.
[NOTE(OS): Shouldn't we just update the language that requires the resolver to be so strict and finally be done with this requirement? Or just give a recommendation in the paragraph on resolver here?]
One issue that has been identified is that not all commonly-used signing software releases include support for an algorithm rollover. This software would need to be updated to support rolling an algorithm before any new algorithms could be deployed.
The registry for a top-level domain (TLD) needs to accept DS records using the new cryptographic algorithm.
Observations to date have shown that some registries only accept DS records with certain algorithms. Registry representatives have indicated that they verify the accuracy of DS records to reduce technical support incidents and ensure customers do not mistakenly create any outages.
However, this means that registries who perform this level of checking must be able to understand new algorithms in order to successfully verify the DS records.
Separately, feedback from registrars has indicated that they do not currently have any mechanism to understand what DNSSEC algorithms a registry can accept.
Registrars perform a critical role in the DNSSEC "chain of trust" of passing the DS record up to the Registry to ensure that the signed zone can be authenticated from the root of DNS all the way to the zone.
If the registrar is also providing the DNS hosting services for a domain, the registrar can easily create the DS record from the DNSKEY record and pass the DS record up to the registry.
However, if the authoritative servers for a domain are not with the registrar, then the registrar needs to provide some mechanism to accept a DS record to pass that up to the registry. Typically this is done through a web interface.
An issue is that many registrar web interfaces only allow the input of DS records using a listed set of DNSSEC algorithms. Any new cryptographic algorithms need to be added to the web interface in order to be accepted into the registrar's system.
Additionally, in a manner similar to registries, many registrars perform some level of verification on the DS record to ensure it was entered "correctly". To do this verification, the registrar's software needs to understand the algorithm used in the DS record. This requires the software to be updated to support the new algorithm.
Note that work is currently underway in {{?I-D.ietf-dnsop-maintain-ds}} to provide an automated mechanism to update the DS records with a registry. If this method becomes widely adopted, registrar web interfaces may no longer be needed.
DNS hosting operators are entities that are operating the authoritative DNS servers for domains and with DNSSEC are also providing the signing of zones. In many cases they may also be the registrar for domain names, but in other cases they are a separate entity providing DNS services to customers.
DNS hosting operators need to update their authoritative DNS server software to understand new cryptographic algorithms, but they also need to update their web interfaces and provisioning software to allow configuration and support of new algorithms.
Beyond the recursive resolvers, authoritative servers, web interfaces and provisioning software, it has been observed that some applications (or "apps"), particularly in the mobile environment, are including their own DNS resolvers within the app itself. These recursive resolvers are used by the app instead of the recursive resolver included with the underlying operating system. These applications that perform DNSSEC validation would need to also be updated to understand a new algorithm.
In many cases, it may be that an underlying developer library needs to be updated first. It will then depend upon how long it takes the application developer to pull in the updated library.
Outside of applications, these developer libraries are also typically used by recursive resolver software and signing software.
This document provides a view into the steps necessary for the deployment of new cryptographic algorithms in DNSSEC at the time of this publication. In order to more rapidly roll out new DNSSEC algorithms, these steps must be understood and hopefully improved over time.
It should be noted that a common theme to emerge from all discussions is a general reluctance to update or change any DNS-related software. "If it isn't broken, don't fix it" is a common refrain. While perhaps understandable from a stability point of view, this attitude creates a challenge for deploying new algorithms.
One potential idea suggested during discussions was for some kind of web-based testing tool that could assist people in understanding what algorithms are supported by different servers and sites.
It is also quite clear that any deployment of new algorithms for DNSSEC use will take a few years to propagate throughout the infrastructure. This needs to be factored in as new algorithms are proposed.
This document does not make any requests of IANA.
No new security considerations are created by this document.
It should be noted that there are security considerations regarding changing DNSSEC algorithms that are mentioned in both {{?RFC6781}} and {{?RFC7583}}.
The information in this document evolved out of several mailing list discussions and also through engagement with participants in the following sessions or events:
The authors thank the participants of the various sessions for their feedback.
NOTE TO RFC EDITOR - Please remove this "Changes" section prior to publication. Thank you.