ENUM L. Conroy
Internet-Draft RMRL
Expires: December 28, 2006 J. Reid
DNS-MODA
June 26, 2006
ENUM Requirement for EDNS0 Support
<draft-conroy-enum-edns0-02.txt>
Status of this Memo
By submitting this Internet-Draft, each author represents that any
applicable patent or other IPR claims of which he or she is aware
have been or will be disclosed, and any of which he or she becomes
aware will be disclosed, in accordance with Section 6 of BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. Note that
other groups may also distribute working documents as Internet-
Drafts.
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."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt.
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html.
This Internet-Draft will expire on December 28, 2006.
Copyright Notice
Copyright (C) The Internet Society (2006).
Abstract
Support for EDNS0 (Extension Mechanisms for DNS) is mandated in this
document for DNS entities querying for or serving NAPTR records. In
general those entities will be supporting ENUM resolution. This
requirement is needed because DNS responses to ENUM-related queries
generally return large RRSets. Without EDNS0 support these lookups
would result in truncated responses and repeated queries over TCP
transport. That has a severe impact on DNS server load and on the
Conroy & Reid Expires December 28, 2006 [Page 1]
Internet-Draft ENUM EDNS0 Requirements June 2006
latency of those queries.
This document updates RFC3761 only in adding this requirement.
Table of Contents
1. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 3
2. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1. DNS - Background . . . . . . . . . . . . . . . . . . . . . 4
3. Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
4. Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
4.1. Required Aspects of EDNS0 Support . . . . . . . . . . . . 7
4.1.1. TCP Requirement . . . . . . . . . . . . . . . . . . . 8
4.1.2. Fragmentation Requirement . . . . . . . . . . . . . . 9
4.1.3. Intermediary Node Requirement . . . . . . . . . . . . 9
5. Security Considerations . . . . . . . . . . . . . . . . . . . 10
6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 11
7. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 12
8. References . . . . . . . . . . . . . . . . . . . . . . . . . . 13
8.1. Normative References . . . . . . . . . . . . . . . . . . . 13
8.2. Informative References . . . . . . . . . . . . . . . . . . 13
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 15
Intellectual Property and Copyright Statements . . . . . . . . . . 16
Conroy & Reid Expires December 28, 2006 [Page 2]
Internet-Draft ENUM EDNS0 Requirements June 2006
1. Terminology
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 BCP 14, RFC2119 [12].
Conroy & Reid Expires December 28, 2006 [Page 3]
Internet-Draft ENUM EDNS0 Requirements June 2006
2. Introduction
ENUM is defined in RFC3761[1]. It uses the underlying DNS protocol
to handle its queries and responses for NAPTR resource records
(defined in [2]) that are to be processed by an ENUM client.
The DNS protocol is defined in RFC1034[3], RFC1035[4] and clarified
in RFC2181[5]. Requirements for Internet Hosts are specified in
RFC1123[6]. DNS is a simple and efficient protocol and is
fundamental to the operation of Internet communications.
Entities involved in processing ENUM queries and responses have to
deal with messages that typically return large sets of resource
records (RRSets). These messages do not fit the profile for which
DNS was originally designed, and so it is necessary to implement the
standard Extension Mechanisms for DNS as described in RFC2671[7],
specifically the feature by which a DNS entity can indicate its
ability to process messages of a given size over UDP transport.
2.1. DNS - Background
For historical reasons a size limit of 512 bytes is specified in
RFC1035 for all messages exchanged in DNS over UDP transport. Small
MTUs were common in early networks and fragmentation issues were not
well addressed in the communications software that existed when the
DNS was introduced. This 512 byte limit was chosen to avoid the risk
of packet fragmentation over paths with a small MTU. When an answer
will not fit within this limit, the DNS response will be truncated
(indicated by the "TC" flag being set to '1' in the response).
DNS queries and responses can also be carried over TCP transport. In
this case, the size limit is not applied because TCP already has
robust mechanisms for handling fragmentation and the reconstruction
of packets. TCP does have performance implications for simple query-
response interactions: for instance by increasing the overall time
taken to complete the transaction and increasing the volume of
network traffic. Thus it is not the default choice of transport for
the DNS protocol.
RFC1035 mandates support for UDP-based queries but only recommends
support for TCP-based queries. However RFC1123 essentially makes TCP
query support mandatory. It mandates that a DNS resolver discard a
truncated response and retry using another transport protocol. In
effect, authoritative name servers that do not answer TCP queries
after returning truncated responses are misconfigured.
With the introduction of the Extension mechanisms described in
RFC2671, there is now a mechanism by which a DNS entity can indicate
Conroy & Reid Expires December 28, 2006 [Page 4]
Internet-Draft ENUM EDNS0 Requirements June 2006
that it is capable of handling UDP-based DNS transactions larger than
those described in RFC1035.
Conroy & Reid Expires December 28, 2006 [Page 5]
Internet-Draft ENUM EDNS0 Requirements June 2006
3. Problem
ENUM zones typically store large sets of resource records (RRSets).
An entry for one E.164 telephone number (i.e. one owner-name) could
contain 10-20 NAPTR records or more. An answer returning such an
RRSet will almost certainly exceed the capacity of a DNS response
meeting the size limit set in RFC1035 for messages using UDP
transport. RFC 1035 (and RFC1123) outline a "fallback" mechanism.
The server indicates that it cannot return the full answer by setting
the TC flag in its response. On receiving this message the client
will discard the partial result and retry the query using TCP
transport.
This fallback induces extra latency and network traffic when
resolving ENUM queries. The initial truncated response is returned
over UDP and discarded, a TCP transport connection is initiated, the
query is repeated and the TCP connection torn down once the complete
answer has been received. These overheads are unacceptable in some
environments where ENUM will be used: high-latency mobile data
networks for instance.
This behaviour also causes extra load on the name servers. They have
to process the initial query and construct a truncated response, only
to receive the query again using TCP transport. Furthermore, even
after it has returned the full answer over a TCP connection the name
server must maintain a TCP control block for a certain time after it
has sent the answer and shutdown of the TCP connection has been
initiated. Answering a high volume of queries using short-lived TCP
connections causes issues with memory usage, involves the name server
in unnecessary processing and may constrict the number of concurrent
connections that may be open. On busy name servers this has severe
operational impact on throughput.
The proportion of conventional DNS queries that exceed the UDP size
limit specified in RFC1035 is relatively small. So the impact on
normal query resolution of this TCP fallback mechanism is minimal.
It just does not happen often enough to be a significant concern.
However for ENUM lookups for NAPTR records this assumption no longer
holds. This fallback procedure will no longer be the exception. It
may well be the norm and performance when handling ENUM queries will
suffer as a result.
Conroy & Reid Expires December 28, 2006 [Page 6]
Internet-Draft ENUM EDNS0 Requirements June 2006
4. Solution
In short, the solution to the problem of returning the large RRSets
typical of ENUM queries is to use EDNS0. This will maintain high
performance and avoid excessive load on DNS servers. An ENUM client
and any resolving name server can use EDNS0 to indicate the size of
UDP packet it is prepared to handle in a DNS response. This allows
name servers involved in the resolution to return answers using UDP
that fit within the limit set by the resolver rather than that
specified in RFC1035. For a description of other situations in which
EDNS0 is useful and for further motivations on its use, see
RFC3225[8] and RFC3226[13].
As well as using EDNS0, it is necessary to ensure that the buffer
sizes reported are adequate. It should be noted that the penalty of
choosing too low a size for EDNS0 support may be even more severe
that the standard method described in RFC 1035 and RFC 1123. Thus it
is good practice to select a larger size than is likely to be needed,
to counteract that greater cost where fallbacks still occur.
Sections 2.4 and particularly 2.5 of [9] explain the rationale for
using the size option of EDNS0 for queries that return larger
responses. In that document, section 3.1 describes expected server
behaviour, section 4.1 describes expected resolver behaviour, while
section 3 summarises the proposed message sizes to be supported by
servers and resolvers. These same size recommendations are repeated
here, as it is felt that ENUM already has a similar issue with larger
responses, and will certainly need the larger messages sizes with the
introduction of IPv6 and DNSSEC support.
4.1. Required Aspects of EDNS0 Support
There are some subtleties with EDNS0 support within ENUM, so the full
implications of the requirement of EDNS0 support for ENUM resolution
are explained here.
The basic requirement for EDNS0 support in ENUM entities is in two
parts:
ALL entities involved in querying for or serving NAPTR records
MUST support EDNS0.
ALL entities involved in querying for or serving NAPTR records
MUST be able to support EDNS0 buffer sizes for queries or
responses of at least 1220 bytes, and SHOULD be able to support
buffer sizes of 4000 bytes.
Entities querying for NAPTR records MUST use EDNS0 in their
queries unless they have current knowledge that EDNS0 support is
Conroy & Reid Expires December 28, 2006 [Page 7]
Internet-Draft ENUM EDNS0 Requirements June 2006
not provided at the target of their queries.
Entities looking up NAPTR records MUST advertise a buffer size of
at least 1220 bytes in their queries, and SHOULD advertise a
buffer size of 4000 bytes. Consideration should also be given to
the MTU of the underlying network, less any overhead needed for
lower-level network protocols.
Of course, support is one thing, but use is another. The mandate for
support of EDNS0 when processing ENUM queries does not imply
spontaneous use. The mechanism described in RFC2671 applies. If a
name server receives a query indicating that the client supports
EDNS0, then it replies with an extended response, assuming that name
server supports ENDS0. If it does not receive such an indication,
then it responds with a conventional RFC1035-style reply. Similarly,
resolvers querying for NAPTR records must indicate their ability to
support EDNS0 and larger buffer sizes when they send those lookups
because this is the only way that they will receive such responses.
There are three further aspects to EDNS0 support.
4.1.1. TCP Requirement
Firstly, it is still possible that ENUM-related queries could result
in truncated responses and TCP retries even though an EDNS0-enabled
mechanism is used. A zone could include a larger set of NAPTR
records than will fit into the packet size the client has reported
itself as supporting. If the ENUM client requests all available
resource records for some ENUM zone rather than just its NAPTR
records, there may be large amounts of data for other resource record
types for the queried owner-name: eg TXT records. In this case the
complete answer may well exceed the client's advertised packet size
even though a NAPTR-specific query would not. Also, the EDNS0 query
may fail for the reasons covered below. In all these cases the
fallback mechanism described in RFC2671 will be needed. For the
fallback process to work for large RRSets, entities will need to
support TCP transport even if EDNS0 is disabled or unavailable for
some reason.
Thus:
If an entity involved knows that EDNS0 queries and responses work
in the current ENUM resolution chain, it MUST be willing to
support queries and responses using TCP transport.
Conroy & Reid Expires December 28, 2006 [Page 8]
Internet-Draft ENUM EDNS0 Requirements June 2006
4.1.2. Fragmentation Requirement
Second, a DNS server may receive queries that indicate a given size
of response is acceptable. However, the resolver may be connected
via a network with a lower MTU, in which case the response packet
will undergo fragmentation and reassembly in transit.
Thus, although obvious (and not directly related to its use in
processing ENUM requests), this means that:
A DNS server responding to a query that includes the EDNS0 size
option MUST NOT set the DF (Don't Fragment) bit in the UDP packet
holding its answer.
4.1.3. Intermediary Node Requirement
The final point concerns intermediate nodes. It has been noticed
that some intermediate nodes exhibit overly aggressive behaviour.
Specifically:
Intermediate nodes MUST NOT block or discard valid ENUM queries
and responses that indicate EDNS0 support. In particular,
intermediate packet filters MUST NOT assume that UDP DNS responses
larger than 512 bytes are invalid. These responses are correct
and MUST NOT be intercepted provided they comply with the EDNS0
standard. Such packet discard strategies are in error.
Intermediate nodes MUST NOT block valid DNS queries and responses
sent over TCP transport. It is perfectly reasonable for DNS
queries to be sent over TCP transport.
This last requirement means that intermediary packet filters MUST NOT
simply block all TCP-based DNS traffic.
Conroy & Reid Expires December 28, 2006 [Page 9]
Internet-Draft ENUM EDNS0 Requirements June 2006
5. Security Considerations
This document does appear to introduce any extra security issues over
and above those mentioned in RFC3761 and in RFC2671, as well as those
listed in the thorough analysis of the threats to DNS in RFC3833
[14].
It should be noted that mandating the use of EDNS0 by ENUM-related
entities also facilitates the deployment of Secure DNS, DNSSEC,
currently defined in RFC4033 [10], RFC4034 [11] and RFC4035 [9].
Secure DNS will be necessary to verify the integrity of ENUM
responses. RFC3225 [8] states that clients signal their ability to
handle signed responses via the DO (DNSSEC OK) bit in the EDNS0
header and a name server will not return these unless this bit is
set. So unless EDNS0 is used, ENUM-related entities will be unable
to verify DNSSEC-signed responses from the DNS. Signed replies from
the DNS are also much larger than unsigned ones, which provided an
added incentive to use larger UDP payloads.
Conroy & Reid Expires December 28, 2006 [Page 10]
Internet-Draft ENUM EDNS0 Requirements June 2006
6. IANA Considerations
This document has no IANA requirements.
Conroy & Reid Expires December 28, 2006 [Page 11]
Internet-Draft ENUM EDNS0 Requirements June 2006
7. Acknowledgements
We would like to thank the working group members active on the ENUM
mailing list who engaged in this topic, and the development and
operational teams that collected data confirming the need for this
mandate.
Conroy & Reid Expires December 28, 2006 [Page 12]
Internet-Draft ENUM EDNS0 Requirements June 2006
8. References
8.1. Normative References
[1] Faltstrom, P. and M. Mealling, "The E.164 to Uniform Resource
Identifiers (URI) Dynamic Delegation Discovery System (DDDS)
Application (ENUM)", RFC 3761, April 2004.
[2] Mealling, M., "Dynamic Delegation Discovery System (DDDS) Part
Three: The Domain Name System (DNS) Database", RFC 3403,
October 2002.
[3] Mockapetris, P., "DOMAIN NAMES - CONCEPTS AND FACILITIES",
RFC 1034, November 1987.
[4] Mockapetris, P., "Domain names - implementation and
specification", STD 13, RFC 1035, November 1987.
[5] Elz, R. and R. Bush, "Clarifications to the DNS Specification",
RFC 2181, July 1997.
[6] Braden, R., "Requirements for Internet Hosts -- Application and
Support", RFC 1123, October 1989.
[7] Vixie, P., "Extension Mechanisms for DNS (EDNS0)", RFC 2671,
August 1999.
[8] Conrad, D., "Indicating Resolver Support of DNSSEC", RFC 3225,
December 2001.
[9] Arends, R., Austein, R., Larson, M., Massey, D., and S. Rose,
"Protocol Modifications for the DNS Security Extensions",
RFC 4035, March 2005.
[10] Arends, R., Austein, R., Larson, M., Massey, D., and S. Rose,
"DNS Security Introduction and Requirements", RFC 4033,
March 2005.
[11] Arends, R., Austein, R., Larson, M., Massey, D., and S. Rose,
"Resource Records for the DNS Security Extensions", RFC 4034,
March 2005.
8.2. Informative References
[12] Bradner, S., "Key words for use in RFCs to Indicate Requirement
Levels", RFC 2119, BCP 14, March 1997.
[13] Gudmundsson, O., "DNSSEC and IPv6 A6 Requirements", RFC 3226,
Conroy & Reid Expires December 28, 2006 [Page 13]
Internet-Draft ENUM EDNS0 Requirements June 2006
December 2001.
[14] Atkins, D. and R. Austein, "Threat Analysis of the Domain Name
System (DNS)", RFC 3833, August 2004.
[15] Bradner, S., "The Internet Standards Process -- Revision 3",
RFC 2026, BCP 9, October 1996.
[16] Bradner, S., "IETF Rights in Contributions", BCP 78, RFC 3978,
March 2005.
[17] Bradner, S., "Intellectual Property Rights in IETF Technology",
BCP 79, RFC 3979, March 2005.
Conroy & Reid Expires December 28, 2006 [Page 14]
Internet-Draft ENUM EDNS0 Requirements June 2006
Authors' Addresses
Lawrence Conroy
Roke Manor Research
Roke Manor
Old Salisbury Lane
Romsey
United Kingdom
Phone: +44 1794 833666
Email: lconroy@insensate.co.uk
URI: http://www.sienum.co.uk
Jim Reid
DNS-MODA
DNS-MODA
6 Langside Court
Bothwell, SCOTLAND
United Kingdom
Phone: +44 1698 852881
Email: jim@dns-moda.org
Conroy & Reid Expires December 28, 2006 [Page 15]
Internet-Draft ENUM EDNS0 Requirements June 2006
Intellectual Property Statement
The IETF takes no position regarding the validity or scope of any
Intellectual Property Rights or other rights that might be claimed to
pertain to the implementation or use of the technology described in
this document or the extent to which any license under such rights
might or might not be available; nor does it represent that it has
made any independent effort to identify any such rights. Information
on the procedures with respect to rights in RFC documents can be
found in BCP 78 and BCP 79.
Copies of IPR disclosures made to the IETF Secretariat and any
assurances of licenses to be made available, or the result of an
attempt made to obtain a general license or permission for the use of
such proprietary rights by implementers or users of this
specification can be obtained from the IETF on-line IPR repository at
http://www.ietf.org/ipr.
The IETF invites any interested party to bring to its attention any
copyrights, patents or patent applications, or other proprietary
rights that may cover technology that may be required to implement
this standard. Please address the information to the IETF at
ietf-ipr@ietf.org.
Disclaimer of Validity
This document and the information contained herein are provided on an
"AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS
OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY AND THE INTERNET
ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED,
INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE
INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED
WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
Copyright Statement
Copyright (C) The Internet Society (2006). This document is subject
to the rights, licenses and restrictions contained in BCP 78, and
except as set forth therein, the authors retain all their rights.
Acknowledgment
Funding for the RFC Editor function is currently provided by the
Internet Society.
Conroy & Reid Expires December 28, 2006 [Page 16]