Use Cases and Requirements for DNS-based Authentication of Named Entities (DANE)
draft-ietf-dane-use-cases-02.txt

Abstract

Many current applications use the certificate-based authentication features in TLS to allow clients to verify that a connected server properly represents a desired domain name. Traditionally, this authentication has been based on PKIX trust hierarchies, rooted in well-known CAs, but additional information can be provided via the DNS itself. This document describes a set of use cases in which the DNS and DNSSEC could be used to make assertions that support the TLS authentication process.

Status of this Memo

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 October 31, 2011.

Copyright Notice

Copyright (c) 2011 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.

Table of Contents

• 1. Introduction
• 2. Definitions
• 3. Use Cases
• 3.1. CA Constraints
• 3.2. Certificate Constraints
• 3.3. Domain-Issued Certificates
• 3.4. Delegated Services
• 3.5. Opportunistic Security
• 3.6. Web Services
• 4. Other Requirements
• 5. Acknowledgements
• 6. IANA Considerations
• 7. Security Considerations
• 8. References
• 8.1. Normative References
• 8.2. Informative References
• Author's Address

1. Introduction

Transport-Layer Security or TLS is used as the basis for security features in many modern Internet applications [RFC5246]. It underlies secure HTTP and secure email [RFC2818][RFC2595][RFC3207], and provides hop-by-hop security in real-time multimedia and instant-messaging protocols [RFC3261][RFC6120].

One feature that is common to most uses of TLS is the use of certificates to authenticate domain names for services. The TLS client begins the TLS connection process with the goal of connecting to a server with a specific domain name. (The process of obtaining this domain name is application-specific. It could be entered by a user or found through an automated discovery process, e.g., via an SRV or NAPTR record.) After obtaining the address of the server via an A or AAAA record, the client conducts a TLS handshake with the server, during which the server presents a PKIX certificate for itself [RFC5280]. Based on this certificate, the client decides whether the server properly represents the desired domain name, and thus whether to proceed with the TLS connection or not.

In most current applications, this decision process is based on PKIX validation and application-specific name matching. The client validates that the certificate chains to a trust anchor [RFC5280], and that the desired domain name is contained in the certificate [RFC6125]. Within this framework, bindings between public keys and domain names are asserted by PKIX CAs. Authentication decisions based on these bindings rely on the authority of these CAs.

The DNS is built to provide information about domain names, and with the advent of DNSSEC [RFC1034][RFC4033], it is possible for this information to be provided securely, in the sense that clients can verify that DNS information was provided by the domain owner. The goal of technologies for DNS-based Authentication of Named Entities (DANE) is to use the DNS and DNSSEC to provide additional information to inform the TLS domain authentication process. This document describes a set of use cases that capture specific goals for using the DNS in this way, and a set of requirements that the ultimate DANE mechanism should satisfy.

2. Definitions

This document also makes use of standard PKIX, DNSSEC, and TLS terminology. See RFC 5280 [RFC5280], RFC 4033 [RFC4033], and RFC 5246 [RFC5246], respectively, for these terms.

Note in particular that the term "server" in this document refers to the server role in TLS, rather than to a host. Multiple servers of this type may be co-located on a single physical host, using different ports, and each of these can use different certificates.

3. Use Cases

In this section, we describe the major use cases that the DANE mechanism should support. This list is not intended to represent all possible ways that the DNS can be used to support TLS authentication. Rather it represents the specific cases that comprise the initial goal for DANE.

In the below use cases, we will refer to the following dramatis personae:

Alice

The operator of a TLS-protected service on the host alice.example.com, and administrator of the corresponding DNS zone.

Bob

A client connecting to alice.example.com

Charlie

A well-known CA that issues certificates with domain names as identifiers

Oscar

An outsourcing provider that operates TLS-protected services on behalf of customers

Trent

A CA that issues certificates with domain names as identifiers, but is not generally well-known.

These use cases are framed in terms of adding protections to TLS server certificates, since the use of these certificates to authenticate server domain names is very common. In applications where TLS clients are also identified by domain names (e.g., XMPP server-to-server connections), the same considerations and use cases can also be applied to TLS client certificates.

3.1. CA Constraints

Alice runs a website on alice.example.com and has obtained a certificate from the well-known CA Charlie. She is concerned that other well-known CAs might issue certificates for alice.example.com without her authorization, which clients would accept. Alice would like to provide a mechanism for visitors to her site to know that they should expect alice.example.com to use a certificate issued under the CA that she uses (Charlie) and not another CA. In TLS terms, Alice is letting Bob know that Charlie's certificate must appear somewhere in the server Certificate message's certificate_list structure.

When Bob connects to alice.example.com, he uses this mechanism to verify that that the certificate presented by the server was issued under the proper CA, Charlie. Bob also performs the normal PKIX validation procedure for this certificate, in particular verifying that the certificate chains to a trust anchor.

Because these constraints do not increase the scope of PKIX-based assertions about domains, there is not a strict requirement for DNSSEC. Deletion of records removes the protection provided by this constraint, but the client is still protected by CA practices (as now). Injected or modified false records are not useful unless the attacker can also obtain a certificate for the target domain. In the worst case, tampering with these constraints increases the risk of false authentication to the level that is now standard.

Injected or modified false records can be used for denial of service, even if the attacker does not have a certificate for the target domain. If an attacker can modify DNS responses that a target host receives, however, there are already much simpler ways of denying service, such as providing a false A or AAAA record. In this case, DNSSEC is not helpful, since an attacker could still case a denial of service by blocking all DNS responses for the target domain.

Continuing to require PKIX validation also limits the degree to which DNS operators (as distinct from the owners of domains) can interfere with TLS authentication through this mechanism. As above, even if a DNS operator falsifies DANE records, it cannot masquerade as the target server unless it can also obtain a certificate for the target domain.

3.2. Certificate Constraints

Alice runs a website on alice.example.com and has obtained a certificate from the well-known CA Charlie. She is concerned about additional, unauthorized certificates being issued by Charlie as well as by other CAs. She would like to provide a way for visitors to her site to know that they should expect alice.example.com to present the specific certificate issued by Charlie. In TLS terms, Alice is letting Bob know that this specific certificate must be the first certificate in the server Certificate message's certificate_list structure.

When Bob connects to alice.example.com, he uses this mechanism to verify that that the certificate presented by the server is the correct certificate. Bob also performs the normal PKIX validation procedure for this certificate, in particular verifying that the certificate chains to a trust anchor.

As in Section 3.1., Alice's assertions about server certificates can be used to constrain the behavior of an outsourcing provider Oscar as well as the CA Charlie and other CAs. Such a certificate constraint requires Oscar to present the specified certificate to clients and not another.

The other security considerations for this case are the same as for the "CA Constraints" case above.

3.3. Domain-Issued Certificates

Alice would like to be able to use generate and use certificates for her website on alice.example.com without involving an external CA at all. Alice can generate her own certificates today, making self-signed certificates and possibly certificates subordinate to those certificates. When Bob receives such a certificate, however, he doesn't have a way to verify that the issuer of the certificate is actually Alice. This concerns him because an attacker could present a different certificate and perform a man in the middle attack. Bob would like to protect against this.

Alice would thus like to have a mechanism for visitors to her site to know that the certificates she issues are actually hers. When Bob connects to alice.example.com, he uses this mechanism to verify that the certificate presented by the server was issued by Alice. Since Bob can bind certificates to Alice in this way, he can use Alice's CA as a trust anchor for purposes of validating certificates for alice.example.com. Alice can additionally recommend that clients accept only her certificates using the CA constraints described above.

This use case is functionally equivalent to the case where Alice doesn't issue her own certificates, but uses a CA Trent that is not well-known. In this case, Alice would be advising Bob that he should treat Trent as a trust anchor for purposes of validating Alice's certificates, rather than a CA operated by Alice herself.

Alice's advertising of trust anchor material in this way does not guarantee that Bob will accept the advertised trust anchor. For example, Bob might have out-of-band information (such as a pre-existing local policy) that indicates that the CA Trent advertised by Alice is not trustworthy, which would lead him to decide not to accept Trent as a TA, and thus to reject Alice's certificate if it is issued under Trent.

Providing trust anchor material in this way clearly requires DNSSEC, since corrupted or injected records could be used by an attacker to cause clients to trust an attacker's certificate. Deleted records will only result in connection failure and denial of service, although this could result in clients re-connecting without TLS (a downgrade attack), depending on the application. Therefore, in order for this use case to be safe, applications must forbid clients from falling back to unsecured channels when records appear to have been deleted (e.g., when a missing record has no NSEC or NSEC3 record).

By the same token, this use case puts the most power in the hands of DNS operators. Since the operator of the appropriate DNS zone has de facto control over the content and signing of the zone, he can create false DANE records that bind a malicious party's certificate to a domain. This risk is especially important to keep in mind in cases where the operator of a DNS zone is a different entity than the owner of the domain, as in DNS hosting/outsourcing arrangements, since in these cases the DNS operator might be able to make changes to a domain that are not authorized by the owner of the domain.

This is not a significant incremental risk, however, relative to the current PKIX-based system. In the current system, CAs need to verify that an entity requesting a certificate for a domain is actually the legitimate holder of that domain. Typically this is done using information published about that domain, such as WHOIS email addresses or special records inserted into a domain. By manipulating these values, it is possible for DNS operators to obtain certificates from some well-known certificate authorities today without authorization from the true domain owner.

3.4. Delegated Services

In addition to guarding against CA mis-issue, CA constraints and certificate constraints can also be used to constrain the set of certificates that can be used by an outsourcing provider. Suppose that Oscar operates alice.example.com on behalf of Alice. In particular, Oscar then has de facto control over what certificates to present in TLS handshakes for alice.example.com. In such cases, there are few ways that DNS-based information about TLS certificates could be configured, for example:

1. Alice has the A/AAAA records in her DNS and can sign them along with the DANE record, but Oscar and Alice now need to have tight coordination if the addresses and/or the certificates change.
2. Alice refers to Oscar's DNS by delegating a sub-domain name to Oscar, and has no control over the A/AAAA, DANE or any other pieces under Oscar's control.
3. Alice can put DANE records into her DNS server, but delegate the address records to Diane's DNS server. This means that Alice can control the usage of certificates but Diane is free to move the servers around as needed. The only coordination needed is when the certificates change, and then it would depend on how the DANE record is setup (i.e. a CA or an EE certificate pointer).

Which of these deployment patterns is used in a given deployment will determine what sort of constraints can be made. In cases where Alice controls DANE records (1 and 3), she can use CA and certificate constraints to control what certificates Oscar presents for Alice's services. For instance, Alice might require Oscar to use certificates under a given set of CAs. This control, however, requires that Alice update DANE records when Oscar needs to change certificates. Cases where Oscar controls DANE records allow Oscar to maintain more autonomy from Alice, but by the same token, Alice cannot make any requirements on the certificates that Oscar uses.

3.5. Opportunistic Security

Alice would like to to publish a web site so that Bob will always have the benefit of the best security his client is capable of, without resulting in a negative user experience when using a legacy browser. For example, suppose that Bob uses two browsers on different machines, one is a legacy browser that does not support DANE and cannot be updated, the other is a browser that has full support for DANE. In this case, the legacy browser should continue to work as before, while the new browser should be able to discover DANE support. In general, the DANE mechanism must allow a clients to determine whether DANE security is available for a site.

3.6. Web Services

A web service is an HTTP-based Internet protocol designed to support direct machine-to-machine communication without the intervention of a human operator or other form of supervisor. Since web services are application protocols, the one aspect of Internet architecture that is essential as far as a Web Service is concerned is that the DNS be used as the naming system for service discovery. Web Services typically evolve over time. A service provider must frequently support legacy clients alongside new and in many cases multiple versions of each protocol. Discovering the certificates or keys to be used to secure the connection to the Web service represents merely one aspect of the more general problem of Web Service property discovery.

4. Other Requirements

In addition to supporting the above use cases, the DANE mechanism must satisfy several lower-level operational and protocol requirements and goals.

Multiple Ports:

DANE should be able to support multiple services with different credentials on the same named host, distinguished by port number.

No Downgrade:

An attacker who can tamper with DNS responses must not be able to make a DANE-compliant client treat a site that has deployed DANE and DNSSEC like a site that has deployed neither.

Encapsulation:

If there is a DANE information for the name alice.example.com, it must only affect services hosted at alice.example.com.

Predictability:

Client behavior in response to DANE information must be spelled out in the DANE specification as precisely as possible, especially for cases where DANE information might conflict with PKIX information.

Simple Key Management:

DANE should have a mode in which the domain owner only needs to maintain a single long-lived public/private key pair.

Minimal Dependencies:

It should be possible for a site to deploy DANE without also deploying anything else, except DNSSEC.

Minimal Options:

Ideally, DANE should have only one operating mode. Practically, DANE should have as few operating modes as possible.

Wild Cards and CNAME:

The mechanism for distributing DANE information should be compatible with the use of DNS wild cards and CNAME records for setting default properties for domains and redirecting services.

5. Acknowledgements

Thanks to Eric Rescorla for the initial formulation of the use cases, Zack Weinberg and Phillip Hallam-Baker for contributing other requirements, and the whole DANE working group for helpful comments on the mailing list.

6. IANA Considerations

This document makes no request of IANA.

7. Security Considerations

The primary focus of this document is the enhancement of TLS authentication procedures using the DNS. The general effect of such mechanisms is to increase the role of DNS operators in authentication processes, either in place of or in addition to traditional third-party actors such as commercial certificate authorities. The specific security implications of the respective use cases are discussed in their respective sections above.

8. References

8.1. Normative References

8.2. Informative References

Author's Address