Internet Engineering Task Force (IETF) T. Reddy Request for Comments: 8094 Cisco Category: Experimental D. Wing ISSN: 2070-1721 P. Patil Cisco February 2017
DNS over Datagram Transport Layer Security (DTLS)
DNS queries and responses are visible to network elements on the path between the DNS client and its server. These queries and responses can contain privacy-sensitive information, which is valuable to protect.
This document proposes the use of Datagram Transport Layer Security (DTLS) for DNS, to protect against passive listeners and certain active attacks. As latency is critical for DNS, this proposal also discusses mechanisms to reduce DTLS round trips and reduce the DTLS handshake size. The proposed mechanism runs over port 853.
Status of This Memo
This document is not an Internet Standards Track specification; it is published for examination, experimental implementation, and evaluation.
This document defines an Experimental Protocol for the Internet community. This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Not all documents approved by the IESG are a candidate for any level of Internet Standard; see Section 2 of RFC 7841.
Copyright (c) 2017 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 ....................................................3 1.1. Relationship to TCP Queries and to DNSSEC ..................3 1.2. Document Status ............................................4 2. Terminology .....................................................4 3. Establishing and Managing DNS over DTLS Sessions ................5 3.1. Session Initiation .........................................5 3.2. DTLS Handshake and Authentication ..........................5 3.3. Established Sessions .......................................6 4. Performance Considerations ......................................7 5. Path MTU (PMTU) Issues ..........................................7 6. Anycast .........................................................8 7. Usage ...........................................................9 8. IANA Considerations .............................................9 9. Security Considerations .........................................9 10. References ....................................................10 10.1. Normative References .....................................10 10.2. Informative References ...................................11 Acknowledgements ..................................................13 Authors' Addresses ................................................13
The Domain Name System is specified in [RFC1034] and [RFC1035]. DNS queries and responses are normally exchanged unencrypted; thus, they are vulnerable to eavesdropping. Such eavesdropping can result in an undesired entity learning domain that a host wishes to access, thus resulting in privacy leakage. The DNS privacy problem is further discussed in [RFC7626].
This document defines DNS over DTLS, which provides confidential DNS communication between stub resolvers and recursive resolvers, stub resolvers and forwarders, and forwarders and recursive resolvers. DNS over DTLS puts an additional computational load on servers. The largest gain for privacy is to protect the communication between the DNS client (the end user's machine) and its caching resolver. DNS over DTLS might work equally between recursive clients and authoritative servers, but this application of the protocol is out of scope for the DNS PRIVate Exchange (DPRIVE) working group per its current charter. This document does not change the format of DNS messages.
The motivations for proposing DNS over DTLS are that:
o TCP suffers from network head-of-line blocking, where the loss of a packet causes all other TCP segments not to be delivered to the application until the lost packet is retransmitted. DNS over DTLS, because it uses UDP, does not suffer from network head-of- line blocking.
o DTLS session resumption consumes one round trip, whereas TLS session resumption can start only after the TCP handshake is complete. However, with TCP Fast Open [RFC7413], the implementation can achieve the same RTT efficiency as DTLS.
Note: DNS over DTLS is an experimental update to DNS, and the experiment will be concluded when the specification is evaluated through implementations and interoperability testing.
DNS queries can be sent over UDP or TCP. The scope of this document, however, is only UDP. DNS over TCP can be protected with TLS, as described in [RFC7858]. DNS over DTLS alone cannot provide privacy for DNS messages in all circumstances, specifically when the DTLS record size is larger than the path MTU. In such situations, the DNS server will respond with a truncated response (see Section 5).
Reddy, et al. Experimental [Page 3]
RFC 8094 DNS over DTLS February 2017
Therefore, DNS clients and servers that implement DNS over DTLS MUST also implement DNS over TLS in order to provide privacy for clients that desire Strict Privacy as described in [DTLS].
DNS Security Extensions (DNSSEC) [RFC4033] provide object integrity of DNS resource records, allowing end users (or their resolver) to verify the legitimacy of responses. However, DNSSEC does not provide privacy for DNS requests or responses. DNS over DTLS works in conjunction with DNSSEC, but DNS over DTLS does not replace the need or value of DNSSEC.
This document is an Experimental RFC. One key aspect to judge whether the approach is usable on a large scale is by observing the uptake, usability, and operational behavior of the protocol in large- scale, real-life deployments.
This DTLS solution was considered by the DPRIVE working group as an option to use in case the TLS-based approach specified in [RFC7858] turns out to have some issues when deployed. At the time of writing, it is expected that [RFC7858] is what will be deployed, and so this specification is mainly intended as a backup.
The following guidelines should be considered when performance benchmarking DNS over DTLS:
1. DNS over DTLS can recover from packet loss and reordering, and does not suffer from network head-of-line blocking. DNS over DTLS performance, in comparison with DNS over TLS, may be better in lossy networks.
2. The number of round trips to send the first DNS query over DNS over DTLS is less than the number of round trips to send the first DNS query over TLS. Even if TCP Fast Open is used, it only works for subsequent TCP connections between the DNS client and server (Section 3 in [RFC7413]).
3. If the DTLS 1.3 protocol [DTLS13] is used for DNS over DTLS, it provides critical latency improvements for connection establishment over DTLS 1.2.
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 [RFC2119] .
Reddy, et al. Experimental [Page 4]
RFC 8094 DNS over DTLS February 2017
3. Establishing and Managing DNS over DTLS Sessions
By default, DNS over DTLS MUST run over standard UDP port 853 as defined in Section 8, unless the DNS server has mutual agreement with its clients to use a port other than 853 for DNS over DTLS. In order to use a port other than 853, both clients and servers would need a configuration option in their software.
The DNS client should determine if the DNS server supports DNS over DTLS by sending a DTLS ClientHello message to port 853 on the server, unless it has mutual agreement with its server to use a port other than port 853 for DNS over DTLS. Such another port MUST NOT be port 53 but MAY be from the "first-come, first-served" port range (User Ports [RFC6335], range 1024-49151). This recommendation against the use of port 53 for DNS over DTLS is to avoid complications in selecting use or non-use of DTLS and to reduce risk of downgrade attacks.
A DNS server that does not support DNS over DTLS will not respond to ClientHello messages sent by the client. If no response is received from that server, and the client has no better round-trip estimate, the client SHOULD retransmit the DTLS ClientHello according to Section 126.96.36.199 of [RFC6347]. After 15 seconds, it SHOULD cease attempts to retransmit its ClientHello. The client MAY repeat that procedure to discover if DNS over DTLS service becomes available from the DNS server, but such probing SHOULD NOT be done more frequently than every 24 hours and MUST NOT be done more frequently than every 15 minutes. This mechanism requires no additional signaling between the client and server.
DNS clients and servers MUST NOT use port 853 to transport cleartext DNS messages. DNS clients MUST NOT send and DNS servers MUST NOT respond to cleartext DNS messages on any port used for DNS over DTLS (including, for example, after a failed DTLS handshake). There are significant security issues in mixing protected and unprotected data; therefore, UDP connections on a port designated by a given server for DNS over DTLS are reserved purely for encrypted communications.
The DNS client initiates the DTLS handshake as described in [RFC6347], following the best practices specified in [RFC7525]. After DTLS negotiation completes, if the DTLS handshake succeeds according to [RFC6347], the connection will be encrypted and would then be protected from eavesdropping.
Reddy, et al. Experimental [Page 5]
RFC 8094 DNS over DTLS February 2017
DNS privacy requires encrypting the query (and response) from passive attacks. Such encryption typically provides integrity protection as a side effect, which means on-path attackers cannot simply inject bogus DNS responses. However, to provide stronger protection from active attackers pretending to be the server, the server itself needs to be authenticated. To authenticate the server providing DNS privacy, DNS client MUST use the authentication mechanisms discussed in [DTLS]. This document does not propose new ideas for authentication.
In DTLS, all data is protected using the same record encoding and mechanisms. When the mechanism described in this document is in effect, DNS messages are encrypted using the standard DTLS record encoding. When a DTLS user wishes to send a DNS message, the data is delivered to the DTLS implementation as an ordinary application data write (e.g., SSL_write()). A single DTLS session can be used to send multiple DNS requests and receive multiple DNS responses.
To mitigate the risk of unintentional server overload, DNS over DTLS clients MUST take care to minimize the number of concurrent DTLS sessions made to any individual server. For any given client/server interaction, it is RECOMMENDED that there be no more than one DTLS session. Similarly, servers MAY impose limits on the number of concurrent DTLS sessions being handled for any particular client IP address or subnet. These limits SHOULD be much looser than the client guidelines above because the server does not know, for example, if a client IP address belongs to a single client, is representing multiple resolvers on a single machine, or is representing multiple clients behind a device performing Network Address Translation (NAT).
In between normal DNS traffic, while the communication to the DNS server is quiescent, the DNS client MAY want to probe the server using DTLS heartbeat [RFC6520] to ensure it has maintained cryptographic state. Such probes can also keep alive firewall or NAT bindings. This probing reduces the frequency of needing a new handshake when a DNS query needs to be resolved, improving the user experience at the cost of bandwidth and processing time.
A DTLS session is terminated by the receipt of an authenticated message that closes the connection (e.g., a DTLS fatal alert). If the server has lost state, a DTLS handshake needs to be initiated with the server. For the server, to mitigate the risk of unintentional server overload, it is RECOMMENDED that the default DNS over DTLS server application-level idle time be set to several seconds and not set to less than a second, but no particular value is
Reddy, et al. Experimental [Page 6]
RFC 8094 DNS over DTLS February 2017
specified. When no DNS queries have been received from the client after idle timeout, the server MUST send a DTLS fatal alert and then destroy its DTLS state. The DTLS fatal alert packet indicates the server has destroyed its state, signaling to the client that if it wants to send a new DTLS message, it will need to re-establish cryptographic context with the server (via full DTLS handshake or DTLS session resumption). In practice, the idle period can vary dynamically, and servers MAY allow idle connections to remain open for longer periods as resources permit.
The DTLS protocol profile for DNS over DTLS is discussed in Section 11 of [DTLS]. To reduce the number of octets of the DTLS handshake, especially the size of the certificate in the ServerHello (which can be several kilobytes), DNS clients and servers can use raw public keys [RFC7250] or Cached Information Extension [RFC7924]. Cached Information Extension avoids transmitting the server's certificate and certificate chain if the client has cached that information from a previous TLS handshake. TLS False Start [RFC7918] can reduce round trips by allowing the TLS second flight of messages (ChangeCipherSpec) to also contain the (encrypted) DNS query.
It is highly advantageous to avoid server-side DTLS state and reduce the number of new DTLS sessions on the server that can be done with TLS Session Resumption without server state [RFC5077]. This also eliminates a round trip for subsequent DNS over DTLS queries, because with [RFC5077] the DTLS session does not need to be re-established.
Since responses within a DTLS session can arrive out of order, clients MUST match responses to outstanding queries on the same DTLS connection using the DNS Message ID. If the response contains a question section, the client MUST match the QNAME, QCLASS, and QTYPE fields. Failure by clients to properly match responses to outstanding queries can have serious consequences for interoperability (Section 7 of [RFC7766]).
Compared to normal DNS, DTLS adds at least 13 octets of header, plus cipher and authentication overhead to every query and every response. This reduces the size of the DNS payload that can be carried. The DNS client and server MUST support the Extension Mechanisms for DNS (EDNS0) option defined in [RFC6891] so that the DNS client can indicate to the DNS server the maximum DNS response size it can reassemble and deliver in the DNS client's network stack. If the DNS client does set the EDNS0 option defined in [RFC6891], then the maximum DNS response size of 512 bytes plus DTLS overhead will be
Reddy, et al. Experimental [Page 7]
RFC 8094 DNS over DTLS February 2017
well within the Path MTU. If the Path MTU is not known, an IP MTU of 1280 bytes SHOULD be assumed. The client sets its EDNS0 value as if DTLS is not being used. The DNS server MUST ensure that the DNS response size does not exceed the Path MTU, i.e., each DTLS record MUST fit within a single datagram, as required by [RFC6347]. The DNS server must consider the amount of record expansion expected by the DTLS processing when calculating the size of DNS response that fits within the path MTU. The Path MTU MUST be greater than or equal to the DNS response size + DTLS overhead of 13 octets + padding size ([RFC7830]) + authentication overhead of the negotiated DTLS cipher suite + block padding (Section 188.8.131.52 of [RFC6347]). If the DNS server's response were to exceed that calculated value, the server MUST send a response that does fit within that value and sets the TC (truncated) bit. Upon receiving a response with the TC bit set and wanting to receive the entire response, the client behavior is governed by the current Usage Profile [DTLS]. For Strict Privacy, the client MUST only send a new DNS request for the same resource record over an encrypted transport (e.g., DNS over TLS [RFC7858]). Clients using Opportunistic Privacy SHOULD try for the best case (an encrypted and authenticated transport) but MAY fall back to intermediate cases and eventually the worst case scenario (clear text) in order to obtain a response.
DNS servers are often configured with anycast addresses. While the network is stable, packets transmitted from a particular source to an anycast address will reach the same server that has the cryptographic context from the DNS over DTLS handshake. But, when the network configuration or routing changes, a DNS over DTLS packet can be received by a server that does not have the necessary cryptographic context. Clients using DNS over DTLS need to always be prepared to re-initiate the DTLS handshake, and in the worst case this could even happen immediately after re-initiating a new handshake. To encourage the client to initiate a new DTLS handshake, DNS servers SHOULD generate a DTLS fatal alert message in response to receiving a DTLS packet for which the server does not have any cryptographic context. Upon receipt of an unauthenticated DTLS fatal alert, the DTLS client validates the fatal alert is within the replay window (Section 184.108.40.206 of [RFC6347]). It is difficult for the DTLS client to validate that the DTLS fatal alert was generated by the DTLS server in response to a request or was generated by an on- or off- path attacker. Thus, upon receipt of an in-window DTLS fatal alert, the client SHOULD continue retransmitting the DTLS packet (in the event the fatal alert was spoofed), and at the same time it SHOULD initiate DTLS session resumption. When the DTLS client receives an authenticated DNS response from one of those DTLS sessions, the other DTLS session should be terminated.
The interaction between a DNS client and a DNS server requires Datagram Transport Layer Security (DTLS) with a ciphersuite offering confidentiality protection. The guidance given in [RFC7525] MUST be followed to avoid attacks on DTLS. The DNS client SHOULD use the TLS Certificate Status Request extension (Section 8 of [RFC6066]), commonly called "OCSP stapling" to check the revocation status of the public key certificate of the DNS server. OCSP stapling, unlike OCSP [RFC6960], does not suffer from scale and privacy issues. DNS clients keeping track of servers known to support DTLS enables clients to detect downgrade attacks. To interfere with DNS over DTLS, an on- or off-path attacker might send an ICMP message towards the DTLS client or DTLS server. As these ICMP messages cannot be authenticated, all ICMP errors should be treated as soft errors [RFC1122]. If the DNS query was sent over DTLS, then the corresponding DNS response MUST only be accepted if it is received over the same DTLS connection. This behavior mitigates all possible attacks described in Measures for Making DNS More Resilient against Forged Answers [RFC5452]. The security considerations in [RFC6347] and [DTLS] are to be taken into account.
A malicious client might attempt to perform a high number of DTLS handshakes with a server. As the clients are not uniquely identified by the protocol and can be obfuscated with IPv4 address sharing and with IPv6 temporary addresses, a server needs to mitigate the impact of such an attack. Such mitigation might involve rate limiting handshakes from a certain subnet or more advanced DoS/DDoS techniques beyond the scope of this document.
[RFC6335] Cotton, M., Eggert, L., Touch, J., Westerlund, M., and S. Cheshire, "Internet Assigned Numbers Authority (IANA) Procedures for the Management of the Service Name and Transport Protocol Port Number Registry", BCP 165, RFC 6335, DOI 10.17487/RFC6335, August 2011, <http://www.rfc-editor.org/info/rfc6335>.
Thanks to Phil Hedrick for his review comments on TCP and to Josh Littlefield for pointing out DNS over DTLS load on busy servers (most notably root servers). The authors would like to thank Simon Josefsson, Daniel Kahn Gillmor, Bob Harold, Ilari Liusvaara, Sara Dickinson, Christian Huitema, Stephane Bortzmeyer, Alexander Mayrhofer, Allison Mankin, Jouni Korhonen, Stephen Farrell, Mirja Kuehlewind, Benoit Claise, and Geoff Huston for discussions and comments on the design of DNS over DTLS. The authors would like to give special thanks to Sara Dickinson for her help.
Tirumaleswar Reddy Cisco Systems, Inc. Cessna Business Park, Varthur Hobli Sarjapur Marathalli Outer Ring Road Bangalore, Karnataka 560103 India