Requirements for Adaptive DNS Discovery

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 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 6 May 2021.¶

Copyright Notice

Copyright (c) 2020 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.¶

Table of Contents

1. Introduction

Several protocols for protecting DNS traffic with encrypted transports have been defined, such as DNS-over-TLS (DoT) [RFC7858] and DNS-over-HTTPS (DoH) [RFC8484]. Encrypted DNS can provide many security and privacy benefits for network clients.¶

While it is possible for clients to statically configure encrypted DNS resolvers to use, dynamic discovery and provisioning of encrypted resolvers can expand the usefulness and applicability of encrypted DNS to many more use cases.¶

The Adaptive DNS Discovery (ADD) Working Group is chartered to define mechanisms that allow clients to automatically discover and select encrypted DNS resolvers in a wide variety of network environments. This document currently focusses on one common use case, that of discovering the encrypted DNS resolver that corresponds to the Do53 resolver offered by a network. Additional use cases can be added in future versions. As well as describing the use case, it lists requirements that any proposed discovery mechanisms should address. They can do this either by providing a solution, or by explicitly stating why it is not in scope.¶

2. Terminology

This document makes use of the following terms.¶

Encrypted DNS: DNS-over-HTTPS [RFC8484], DNS-over-TLS [RFC7858], or any other encrypted DNS technology that the IETF may publish, such as DNS-over-QUIC [I-D.ietf-dprive-dnsoquic].¶

Do53: Unencrypted DNS over UDP port 53, or TCP port 53 [RFC1035].¶

Equivalent: See Section 3.1.¶

3. Use case description

It is often the case that a client possesses no specific configuration for how to operate DNS, and at some point joins a network that it has no previous knowledge about. In such a case the usual existing behaviour is to dynamically discover the network's recommended Do53 resolver and use it. This long-standing practice works in nearly all networks, but presents a number of privacy and security risks that were the motivation for the development of encrypted DNS.¶

The network's recommended unencrypted resolver may have a number of properties that differ from a generic resolver. It may be able to answer names that are not known globally, it may exclude some names (for positive or negative reasons), and it may provide address answers that have improved proximity. In this use case it is assumed that the user who chose to join this network would also like to make use of these properties of the network's unencrypted resolver, at least some of the time. However they would like to use an encrypted DNS protocol rather than Do53.¶

Using an encrypted and authenticated resolver that is equivalent to the one provisioned by the network can provide several benefits that are not possible if only unencrypted DNS is used:¶

• Prevent other devices on the network from observing client DNS messages¶
• Authenticate that the DNS resolver is the correct one¶
• Verify that answers come from the selected DNS resolver¶

To meet this case there should be a means by which the client can learn how to contact an encrypted DNS resolver that provides equivalent responses as the ones served by the network's recommended unencrypted resolver. It is not a requirement that these two resolvers are the same physical or logical machine. Often they will be, but they could equally be separated, perhaps by hundreds of miles. However it is deployed, the key is that they are equivalent.¶

4. Network-identified encrypted resolvers

DNS servers are often provisioned by a network as part of DHCP options [RFC2132], IPv6 Router Advertisement (RA) options [RFC8106], Point-to-Point Protocol (PPP) [RFC1877], or 3GPP Protocol Configuration Options (TS24.008). Historically this is usually one or more Do53 resolver IP addresses, to be used for traditional unencrypted DNS.¶

A solution is required that enhances the set of information delivered to include details of one or more equivalent encrypted DNS resolvers, or states that there are none.¶

5. Resolver-identified encrypted resolvers

To support cases where the network is unable to identify an encrypted resolver, it should be possible to learn the details of one or more equivalent encrypted DNS resolvers by communicating with the network-recommended unencrypted Do53 resolver. This should involve an exchange that uses standard DNS messages that can be handled, or forwarded, by existing deployed software.¶

It is frequently the case that Do53 resolvers announced by home networks are difficult to upgrade to support encrypted operation. In such cases it is possible that the only option for encrypted operation is to refer to a separate globally-addressed encrypted DNS resolver.¶

If the local resolver has been upgraded to support encrypted DNS, the client may not initially be aware that its local resolver supports it. Discovering this may require communication with the local resolver, or an upstream resolver, over Do53. Clients that choose to use this local encrypted DNS gain the benefits of encryption while retaining the benefits of a local caching resolver with knowledge of the local topology.¶

An additional benefit of using a local resolver occurs with IoT devices. A common usage pattern for such devices is for it to "call home" to a service that resides on the public Internet, where that service is referenced through a domain name. As discussed in Manufacturer Usage Description Specification [RFC8520], because these devices tend to require access to very few sites, all other access should be considered suspect. However, if the query is not accessible for inspection, it becomes quite difficult for the infrastructure to suspect anything.¶

6. Privacy and security requirements

Encrypted (and authenticated) DNS improves the privacy and security of DNS queries and answers in the presence of malicious attackers. Such attackers are assumed to interfere with or otherwise impede DNS traffic and corresponding discovery mechanisms. They may be on-path or off-path between the client and entities with which the client communicates [RFC3552]. These attackers can inject, tamper, or otherwise interfere with traffic as needed. Given these capabilities, an attacker may have a variety of goals, including, though not limited to:¶

• Monitor and profile clients by observing unencrypted DNS traffic¶
• Modify unencrypted DNS traffic to filter or augment the user experience¶
• Block encrypted DNS¶

Given this type of attacker, resolver discovery mechanisms must be designed carefully to not worsen a client's security or privacy posture. In particular, attackers under consideration must not be able to:¶

• Redirect secure DNS traffic to themselves when they would not otherwise handle DNS traffic.¶
• Override or interfere with the resolver preferences of a user or administrator.¶
• Cause clients to use a discovered resolver which has no authenticated delegation from a client-known entity.¶
• Influence automatic discovery mechanisms such that a client uses one or more resolvers that are not otherwise involved with providing service to the client, such as: a network provider, a VPN server, a content provider being accessed, or a server that the client has manually configured.¶

When discovering DNS resolvers on a local network, clients have no mechanism to distinguish between cases where an active attacker with the above capabilities is interfering with discovery, and situations wherein the network has no encrypted resolver. Absent such a mechanism, an attacker can always succeed in these goals. Therefore, in such circumstances, viable solutions for local DNS resolver discovery should consider weaker attackers, such as those with only passive eavesdropping capabilities. It is unknown whether such relaxations represent a realistic attacker in practice. Thus, local discovery solutions designed around this threat model may have limited value.¶

7. Statement of Requirements

This section lists requirements that flow from the above sections.¶

8. Security Considerations

See Section 6.¶

9. IANA Considerations

This document has no IANA actions.¶

Acknowledgments

This document was started based on discussion during the ADD meeting of IETF108, the subsequent interims, on the list, and with text from draft-pauly-add-requirements. In particular this document was informed by contributions from Martin Thomson, Eric Rescorla, Tommy Jensen, Ben Schwartz, Paul Hoffman, Ralf Weber, Michael Richardson, Mohamed Boucadair, Sanjay Mishra, Jim Reid, Neil Cook, Nic Leymann and Andrew Campling.¶

Authors' Addresses