DNS over TCP: DNS Transport over TCP - Implementation Requirements
Why DNS implementations must support TCP in addition to UDP.
Bellis, R., "DNS Transport over TCP - Implementation Requirements", RFC 5966, DOI 10.17487/RFC5966, August 2010, <https://www.rfc-editor.org/info/rfc5966>.
@misc{rfc5966,
series = {Request for Comments},
number = 5966,
howpublished = {RFC 5966},
publisher = {RFC Editor},
doi = {10.17487/RFC5966},
url = {https://www.rfc-editor.org/info/rfc5966},
author = {R. Bellis},
title = {{DNS Transport over TCP - Implementation Requirements}},
pagetotal = 7,
year = 2010,
month = aug,
} RFC text
Contents
Internet Engineering Task Force (IETF) R. Bellis
Request for Comments: 5966 Nominet UK
Updates: 1035, 1123 August 2010
Category: Standards Track
ISSN: 2070-1721
DNS Transport over TCP - Implementation Requirements
Abstract
This document updates the requirements for the support of TCP as a
transport protocol for DNS implementations.
Status of This Memo
This is an Internet Standards Track document.
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). Further information on
Internet Standards is available in Section 2 of RFC 5741.
Information about the current status of this document, any errata,
and how to provide feedback on it may be obtained at
http://www.rfc-editor.org/info/rfc5966.
Copyright Notice
Copyright (c) 2010 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
2. Terminology Used in This Document . . . . . . . . . . . . . . . 3
3. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . 3
4. Transport Protocol Selection . . . . . . . . . . . . . . . . . 4
5. Connection Handling . . . . . . . . . . . . . . . . . . . . . . 5
6. Response Reordering . . . . . . . . . . . . . . . . . . . . . . 6
7. Security Considerations . . . . . . . . . . . . . . . . . . . . 6
8. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 7
9. References . . . . . . . . . . . . . . . . . . . . . . . . . . 7
9.1. Normative References . . . . . . . . . . . . . . . . . . . 7
9.2. Informative References . . . . . . . . . . . . . . . . . . 7
1. Introduction
Most DNS [RFC1034] transactions take place over UDP [RFC0768]. TCP
[RFC0793] is always used for zone transfers and is often used for
messages whose sizes exceed the DNS protocol's original 512-byte
limit.
Section 6.1.3.2 of [RFC1123] states:
DNS resolvers and recursive servers MUST support UDP, and SHOULD
support TCP, for sending (non-zone-transfer) queries.
However, some implementors have taken the text quoted above to mean
that TCP support is an optional feature of the DNS protocol.
The majority of DNS server operators already support TCP and the
default configuration for most software implementations is to support
TCP. The primary audience for this document is those implementors
whose failure to support TCP restricts interoperability and limits
deployment of new DNS features.
This document therefore updates the core DNS protocol specifications
such that support for TCP is henceforth a REQUIRED part of a full DNS
protocol implementation.
Whilst this document makes no specific recommendations to operators
of DNS servers, it should be noted that failure to support TCP (or
the blocking of DNS over TCP at the network layer) may result in
resolution failure and/or application-level timeouts.
2. Terminology Used in This Document
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in [RFC2119].
3. Discussion
In the absence of EDNS0 (Extension Mechanisms for DNS 0) (see below),
the normal behaviour of any DNS server needing to send a UDP response
that would exceed the 512-byte limit is for the server to truncate
the response so that it fits within that limit and then set the TC
flag in the response header. When the client receives such a
response, it takes the TC flag as an indication that it should retry
over TCP instead.
RFC 1123 also says:
... it is also clear that some new DNS record types defined in the
future will contain information exceeding the 512 byte limit that
applies to UDP, and hence will require TCP. Thus, resolvers and
name servers should implement TCP services as a backup to UDP
today, with the knowledge that they will require the TCP service
in the future.
Existing deployments of DNS Security (DNSSEC) [RFC4033] have shown
that truncation at the 512-byte boundary is now commonplace. For
example, a Non-Existent Domain (NXDOMAIN) (RCODE == 3) response from
a DNSSEC-signed zone using NextSECure 3 (NSEC3) [RFC5155] is almost
invariably larger than 512 bytes.
Since the original core specifications for DNS were written, the
Extension Mechanisms for DNS (EDNS0 [RFC2671]) have been introduced.
These extensions can be used to indicate that the client is prepared
to receive UDP responses larger than 512 bytes. An EDNS0-compatible
server receiving a request from an EDNS0-compatible client may send
UDP packets up to that client's announced buffer size without
truncation.
However, transport of UDP packets that exceed the size of the path
MTU causes IP packet fragmentation, which has been found to be
unreliable in some circumstances. Many firewalls routinely block
fragmented IP packets, and some do not implement the algorithms
necessary to reassemble fragmented packets. Worse still, some
network devices deliberately refuse to handle DNS packets containing
EDNS0 options. Other issues relating to UDP transport and packet
size are discussed in [RFC5625].
The MTU most commonly found in the core of the Internet is around
1500 bytes, and even that limit is routinely exceeded by DNSSEC-
signed responses.
The future that was anticipated in RFC 1123 has arrived, and the only
standardised UDP-based mechanism that may have resolved the packet
size issue has been found inadequate.
4. Transport Protocol Selection
All general-purpose DNS implementations MUST support both UDP and TCP
transport.
o Authoritative server implementations MUST support TCP so that they
do not limit the size of responses to what fits in a single UDP
packet.
o Recursive server (or forwarder) implementations MUST support TCP
so that they do not prevent large responses from a TCP-capable
server from reaching its TCP-capable clients.
o Stub resolver implementations (e.g., an operating system's DNS
resolution library) MUST support TCP since to do otherwise would
limit their interoperability with their own clients and with
upstream servers.
Stub resolver implementations MAY omit support for TCP when
specifically designed for deployment in restricted environments where
truncation can never occur or where truncated DNS responses are
acceptable.
Regarding the choice of when to use UDP or TCP, Section 6.1.3.2 of
RFC 1123 also says:
... a DNS resolver or server that is sending a non-zone-transfer
query MUST send a UDP query first.
That requirement is hereby relaxed. A resolver SHOULD send a UDP
query first, but MAY elect to send a TCP query instead if it has good
reason to expect the response would be truncated if it were sent over
UDP (with or without EDNS0) or for other operational reasons, in
particular, if it already has an open TCP connection to the server.
5. Connection Handling
Section 4.2.2 of [RFC1035] says:
If the server needs to close a dormant connection to reclaim
resources, it should wait until the connection has been idle for a
period on the order of two minutes. In particular, the server
should allow the SOA and AXFR request sequence (which begins a
refresh operation) to be made on a single connection. Since the
server would be unable to answer queries anyway, a unilateral
close or reset may be used instead of a graceful close.
Other more modern protocols (e.g., HTTP [RFC2616]) have support for
persistent TCP connections and operational experience has shown that
long timeouts can easily cause resource exhaustion and poor response
under heavy load. Intentionally opening many connections and leaving
them dormant can trivially create a "denial-of-service" attack.
It is therefore RECOMMENDED that the default application-level idle
period should be of the order of seconds, but no particular value is
specified. In practise, the idle period may vary dynamically, and
servers MAY allow dormant connections to remain open for longer
periods as resources permit.
To mitigate the risk of unintentional server overload, DNS clients
MUST take care to minimize the number of concurrent TCP connections
made to any individual server. Similarly, servers MAY impose limits
on the number of concurrent TCP connections being handled for any
particular client.
Further recommendations for the tuning of TCP stacks to allow higher
throughput or improved resiliency against denial-of-service attacks
are outside the scope of this document.
6. Response Reordering
RFC 1035 is ambiguous on the question of whether TCP queries may be
reordered -- the only relevant text is in Section 4.2.1, which
relates to UDP:
Queries or their responses may be reordered by the network, or by
processing in name servers, so resolvers should not depend on them
being returned in order.
For the avoidance of future doubt, this requirement is clarified.
Client resolvers MUST be able to process responses that arrive in a
different order to that in which the requests were sent, regardless
of the transport protocol in use.
7. Security Considerations
Some DNS server operators have expressed concern that wider use of
DNS over TCP will expose them to a higher risk of denial-of-service
(DoS) attacks.
Although there is a higher risk of such attacks against TCP-enabled
servers, techniques for the mitigation of DoS attacks at the network
level have improved substantially since DNS was first designed.
At the time of writing, the vast majority of Top Level Domain (TLD)
authority servers and all of the root name servers support TCP and
the author knows of no evidence to suggest that TCP-based DoS attacks
against existing DNS infrastructure are commonplace.
That notwithstanding, readers are advised to familiarise themselves
with [CPNI-TCP].
Operators of recursive servers should ensure that they only accept
connections from expected clients, and do not accept them from
unknown sources. In the case of UDP traffic, this will help protect
against reflector attacks [RFC5358] and in the case of TCP traffic it
will prevent an unknown client from exhausting the server's limits on
the number of concurrent connections.
8. Acknowledgements
The author would like to thank the document reviewers from the DNSEXT
Working Group, and in particular, George Barwood, Alex Bligh, Alfred
Hoenes, Fernando Gont, Olafur Gudmondsson, Jim Reid, Paul Vixie, and
Nicholas Weaver.
9. References
9.1. Normative References
[RFC0768] Postel, J., "User Datagram Protocol", STD 6, RFC 768,
August 1980.
[RFC0793] Postel, J., "Transmission Control Protocol", STD 7,
RFC 793, September 1981.
[RFC1034] Mockapetris, P., "Domain names - concepts and
facilities", STD 13, RFC 1034, November 1987.
[RFC1035] Mockapetris, P., "Domain names - implementation and
specification", STD 13, RFC 1035, November 1987.
[RFC1123] Braden, R., "Requirements for Internet Hosts -
Application and Support", STD 3, RFC 1123, October 1989.
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119, March 1997.
[RFC2671] Vixie, P., "Extension Mechanisms for DNS (EDNS0)",
RFC 2671, August 1999.
9.2. Informative References
[CPNI-TCP] CPNI, "Security Assessment of the Transmission Control
Protocol (TCP)", 2009, <http://www.cpni.gov.uk/Docs/
tn-03-09-security-assessment-TCP.pdf>.
[RFC2616] Fielding, R., Gettys, J., Mogul, J., Frystyk, H.,
Masinter, L., Leach, P., and T. Berners-Lee, "Hypertext
Transfer Protocol -- HTTP/1.1", RFC 2616, June 1999.
[RFC4033] Arends, R., Austein, R., Larson, M., Massey, D., and S.
Rose, "DNS Security Introduction and Requirements",
RFC 4033, March 2005.
[RFC5155] Laurie, B., Sisson, G., Arends, R., and D. Blacka, "DNS
Security (DNSSEC) Hashed Authenticated Denial of
Existence", RFC 5155, March 2008.
[RFC5358] Damas, J. and F. Neves, "Preventing Use of Recursive
Nameservers in Reflector Attacks", BCP 140, RFC 5358,
October 2008.
[RFC5625] Bellis, R., "DNS Proxy Implementation Guidelines",
BCP 152, RFC 5625, August 2009.
Author's Address
Ray Bellis
Nominet UK
Edmund Halley Road
Oxford OX4 4DQ
United Kingdom
Phone: +44 1865 332211
EMail: ray.bellis@nominet.org.uk
URI: http://www.nominet.org.uk/
AI ELI5
On this page
§1Why TCP Matters for DNS
The DNS protocol works primarily over UDP, a connectionless protocol that sends one message at a time with a maximum size limit. The original DNS specification from 1987 imposed a 512-byte limit on UDP responses. Zone transfers—the process of copying all records from one DNS server to another—have always required TCP because they often exceed this limit.
DNS specifies that when a response would exceed 512 bytes, the server should truncate it and set the TC (truncation) flag in the response header. The client then knows to retry the same query over TCP, which has no size limit. However, treating TCP as optional has caused interoperability problems. This RFC elevates TCP from an optional feature to a mandatory requirement for all DNS implementations.
DNS is like sending a postcard: fast but small. A bigger answer needs a letter (TCP), which takes longer but has no size limit. Old DNS said postcard-first was OK, but now computers must know how to send letters too.
§2The Size Problem and Modern DNS
When DNS Security (DNSSEC) was deployed in real systems, it made responses much larger. For example, an NXDOMAIN response (negative answer indicating a name does not exist) from a DNSSEC-signed zone using NSEC3 (Next Secured record hashing) is almost always larger than 512 bytes.
The DNS protocol supports EDNS0 (Extension Mechanisms for DNS), which allows clients to advertise that they can accept larger UDP responses. However, UDP packets larger than the path MTU (the maximum packet size networks can carry) must be fragmented into smaller pieces. Fragmented DNS packets are unreliable in practice: many firewalls block them, some network devices refuse to handle DNS packets with EDNS0 options, and reassembly often fails.
The core Internet MTU is typically around 1500 bytes, and even that is routinely exceeded by DNSSEC responses. TCP, which breaks data into smaller pieces automatically, is the only reliable way to handle large responses.
Modern DNS with security signatures makes answers huge. Splitting big UDP messages breaks them. The future needs TCP, which automatically chops things into pieces that always arrive.
§3TCP Support is Now Mandatory
All general-purpose DNS implementations MUST support both UDP and TCP transport. Specifically:
- Authoritative servers (the servers that hold the official records) MUST support TCP so they can send responses of any size without truncation.
- Recursive servers (which look up names on behalf of clients) MUST support TCP so they can relay large responses from other servers to their own clients.
- Stub resolvers (the DNS client library on an operating system) MUST support TCP so they do not limit their ability to work with upstream servers.
The only exception is stub resolvers designed for restricted environments where responses are guaranteed never to be truncated.
The original RFC 1123 required that DNS resolvers "SHOULD" send a UDP query first before trying TCP. This RFC relaxes that requirement: resolvers SHOULD send UDP first, but MAY skip directly to TCP if they have good reason to expect truncation or if they already have an open TCP connection to the server.
Every DNS program must know how to use TCP now. Try UDP first (it is faster), but switch to TCP if needed. Old advice only said "should," but now it is the rule.
§4Managing Persistent TCP Connections
The original DNS specification suggested that servers could keep TCP connections open for about two minutes of idle time to allow zone transfer sequences to occur on a single connection. Modern protocols like HTTP have shown that long timeouts cause resource exhaustion and denial-of-service vulnerabilities.
The RFC recommends that the default idle timeout should be on the order of seconds, not minutes, though implementations may vary this dynamically based on load. Clients MUST minimize the number of concurrent TCP connections to any server to avoid overwhelming it. Servers MAY limit the number of concurrent connections from any client.
Keep TCP connections for seconds, not minutes, so the server does not run out of space. Clients should not open tons of connections at once.
§5Response Reordering and Out-of-Order Replies
DNS allows queries and responses to arrive in any order, whether over UDP or TCP. Clients sending multiple queries at once cannot assume the replies arrive in the same order the queries were sent. Client software MUST be prepared to match responses to their corresponding requests by message ID, regardless of order.
When you send several questions, the answers might come back scrambled. Always check which question each answer belongs to, not just the order they arrive.
§6DoS Risks and Mitigation
Some DNS operators worried that TCP support would increase denial-of-service (DoS) attack risk. While TCP connections do have higher attack exposure than stateless UDP, network-level DoS defenses have improved substantially since DNS was designed. Most DNS servers today, including all root servers and most top-level domain (TLD) servers, support TCP and no evidence suggests that TCP-based attacks are widespread.
Operators of recursive servers should accept connections only from expected clients, not from unknown sources, to prevent both UDP reflector attacks and TCP connection exhaustion attacks.
Attackers could create more problems with TCP. But the Internet's defenses are much better now, and almost every major DNS server already uses TCP safely.