Internet DRAFT - draft-wkumari-dnsop-multiple-responses
draft-wkumari-dnsop-multiple-responses
dnsop W. Kumari
Internet-Draft Google
Intended status: Standards Track Z. Yan
Expires: January 4, 2018 CNNIC
W. Hardaker
USC/ISI
D. Lawrence
Akamai Technologies
July 3, 2017
Returning extra answers in DNS responses.
draft-wkumari-dnsop-multiple-responses-05
Abstract
This document (re)introduces the ability to provide multiple answers
in a DNS response. This is especially useful as, in many cases, the
entity making the request has no a prori knowledge of what other
questions it will need to ask.
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 January 4, 2018.
Copyright Notice
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
Kumari, et al. Expires January 4, 2018 [Page 1]
Internet-Draft DNS Extra Answers July 2017
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
1.1. Requirements notation . . . . . . . . . . . . . . . . . . 3
2. Background . . . . . . . . . . . . . . . . . . . . . . . . . 3
3. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 4
4. Returning multiple answers . . . . . . . . . . . . . . . . . 4
5. The EXTRA Resource Record . . . . . . . . . . . . . . . . . . 5
5.1. File Format . . . . . . . . . . . . . . . . . . . . . . . 5
5.2. Wire Format . . . . . . . . . . . . . . . . . . . . . . . 5
6. Signaling support . . . . . . . . . . . . . . . . . . . . . . 6
7. Stub-Resolver Considerations . . . . . . . . . . . . . . . . 6
8. Use of Additional information . . . . . . . . . . . . . . . . 6
9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 6
10. Security Considerations . . . . . . . . . . . . . . . . . . . 7
11. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 7
12. Normative References . . . . . . . . . . . . . . . . . . . . 7
Appendix A. Changes / Author Notes. . . . . . . . . . . . . . . 8
Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 9
1. Introduction
In many cases a name being resolved in the DNS provides the reason
behind why the name is being resolved. This may allow the
authoritative nameserver to predict what other answers a recursive
resolver will soon query for. By providing multiple answers in the
response, the authoritative name server operator can assist a caching
recursive resolver in pre-populating its cache before a stub resolver
or other client asks for the subsequent queries. Apart from
decreasing the latency for end users [RFC6555], this also decreases
the total number of queries that the recursive resolver needs to send
and the authoritative server needs to answer.
For example, the domain name administrator of Example Widgets, Inc
(example.com) knows that the web page at www.example.com contains
various other resources, including some images (served from
images.example.com), some Cascading Style Sheets (served from
css.example.com) and some JavaScript (served from data.example.com).
An application attempting to resolve www.example.com is very likely
to be a web browser rendering the page and will likely also need to
resolve all of these additional names as well. Providing all of
these answers in response to a query for www.example.com allows the
recursive resolver to pre-populate its cache and have these answers
Kumari, et al. Expires January 4, 2018 [Page 2]
Internet-Draft DNS Extra Answers July 2017
available immediately when a stub resolver or other DNS client asks
for them. What is important to notice here is that the stub resolver
does not know what other questions it will need to make until after
it has already made the request for www.exmaple.com, received the
reply, made the HTTP connection and parsed the HTML.
Other examples where this technique may be useful include SMTP (by
including mail server addresses, SPF and DKIM records when serving
the MX record), SRV (by providing the target information in addition
to the SRV response) and TLSA (by providing any TLSA records
associated with a name). This same technique can also be used to
include both the IPv4 (A) and IPv6 (AAAA) addresses for any singular
address query.
This technique, described in this document, is purely an optimization
and enables a zone publisher to distribute other related answers that
the client is likely to need along with an answer to the original
request. Users get a better experience, recursive resolvers need to
send less queries, authoritative servers have to answer fewer
queries, etc.
1.1. Requirements notation
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].
2. Background
The existing DNS specifications [RFC1034] allow for supplemental
information to be included in the "additional" section of the DNS
response, but in order to defeat cache poisoning attacks most
implementations either ignore or don't trust additional records they
didn't ask for. For more background, see [Ref.Bellovin] and
[RFC2181].
Not trusting the information in the additional section was necessary
since there was no way to authenticate it. If a resolver queries for
www.example.com and received answers for www.invalid.com as well, it
is impossible for a non-validating resolver to tell if these were
actually from invalid.com or if an attacker was trying to push bad
information for invalid.com into the resolver's cache. In a world of
ubiquitous DNSSEC deployment [Ed note: By the time this document is
published, there *will* be ubiquitous DNSSEC :-) ], a validating
resolver can validate, authenticate and trust the records in the
additional information.
Kumari, et al. Expires January 4, 2018 [Page 3]
Internet-Draft DNS Extra Answers July 2017
3. Terminology
Additional records Additional records are records that the
authoritative nameserver has included in the Additional section.
EXTRA Resource Record The EXTRA resource record (defined below)
carries a list of additional records to send.
Primary query A Primary query (or primary question) is a QNAME that
the name server operator would like to return additional answers
for.
Supporting DNSSEC information Supporting DNSSEC information is the
DNSSEC RRSIGs that prove the authenticity of the Additional
records.
Stub Resolver The term "Stub Resolver" is used in this document to
refer to the most common instance of a DNS client sending DNS
requests to a Recursive Resolver. However, other DNS clients are
not excluded from these usages and where we write "Stub Resolver"
you may read it as "Stub Resolver or other DNS client".
4. Returning multiple answers
The authoritative nameserver should include as many of the instructed
additional records identified by the Extra Resource Record and
Supporting DNSSEC information as will fit in the response packet.
These additional records (and Supporting DNSSEC information) are
appended to the additional section of the response.
In order to include additional records in a response, these
conditions need to be met:
1. Additional records MUST only be included when the Name Server is
authoritative for the zone, and the records to be returned are
DNSSEC signed.
2. The supporting DNSSEC information necessary to perform validation
on the records MUST be included.
3. The Authoritative Name Server SHOULD include as many of the
additional records as will fit in the response. Additional
records SHOULD be inserted in the order specified in the
Additional records list.
4. Zone administrators SHOULD only include records identified in the
EXTRA Resource Records that they expect a client to need.
Kumari, et al. Expires January 4, 2018 [Page 4]
Internet-Draft DNS Extra Answers July 2017
5. The EXTRA Resource Record
To allow a zone content administrator to instruct the name server
which additional records to serve when it receives a query to a
label, we introduce the EXTRA Resource Record (RR). These additional
records are appended to the additional section (note that the EXTRA
RR itself is not appended). The EXTRA resource record MAY still be
queried for directly (e.g for debugging), in which case the record
itself is returned.
5.1. File Format
The format of the EXTRA RR is:
label EXTRA "label,type; label,type; label,type; ..."
For example, if the operator of example.com would like to also return
A record answers for images.example.com, css.html.example.com and
both an A and AAAA for data.example.com when queried for
www.example.com, they would create the following record:
www.example.com. EXTRA "images,A; css,A; data,A; data,AAAA;"
The entries in the EXTRA list are ordered. An authoritative
nameserver SHOULD insert the records in the order listed when filling
the response packet. This is to allow the operator to express a
preference in case all the records will not fit in the response. The
TTL of the records added to the Additional section are MUST be the
same as if queried directly.
In some cases a zone content administrator might not know what all
additional records clients need. For example, the owner of
www.example.com may have outsourced his DNS operations to a third
party, and / or the DNS operator might not interact with the web
development team. DNS server operators may use tools to mine their
query logs for records to include. For example, if, in a large
majority of cases, a recursive server asks for foo.example.com and
then very soon after asks for bar.example.com, it may make sense to
optimize this by opportunistically returning bar when queried for
foo. This functionality could also be included in the authoritative
name server software itself. The exact mechanisms and heuristics
used for this are not discussed in this document.
5.2. Wire Format
The wire format of the EXTRA RR is the same as the wire format for a
TXT RR:
Kumari, et al. Expires January 4, 2018 [Page 5]
Internet-Draft DNS Extra Answers July 2017
+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
/ TXT-DATA /
+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
Where TXT-DATA is one or more <character-string>s.
The EXTRA RR has RR type TBD [RFC Editor: insert the IANA assigned
value and delete this note]
6. Signaling support
Recursive Resolvers (or other DNS clients) that support EXTRA records
MAY signal this by setting the OPT record's EXTRA bit (bit NN [TBD:
assigned by IANA] in the EDNS0 extension header to 1).
7. Stub-Resolver Considerations
No modifications need to be made to stub-resolvers to get the
predominate benefit of this protocol, since the majority of the speed
gain will take place between the validating recursive resolver and
the authoritative name server. However, stub resolvers may choose to
support this technique, and / or may query directly for the EXTRA RR
if it wants to pre-query for data that will likely be needed in the
process of supporting applications.
8. Use of Additional information
When deciding to use additional records in the additional section, a
resolver must follow certain rules:
1. Additional records MUST be validated before being used.
2. Additional records SHOULD have lower priority in the cache than
answers received because they were requested. This is to help
evict Additional records from the cache first (to help prevent
cache filling attacks).
3. Recursive resolvers MAY choose to ignore Additional records for
any reason, including CPU or cache space concerns, phase of the
moon, etc. It may choose to accept all, some or none of the
Additional records.
9. IANA Considerations
This document contains the following IANA assignment requirements:
1. The EXTRA bit discussed in Section 6 needs to be allocated. [ Ed:
This section to be completed later ]
Kumari, et al. Expires January 4, 2018 [Page 6]
Internet-Draft DNS Extra Answers July 2017
10. Security Considerations
Additional records will make DNS responses even larger than they are
currently, leading to larger records that can be used in DNS
reflection attacks. One could mitigate this by only serving
responses to EXTRA requests over TCP or when using Cookies [RFC5395],
although there is no easy way to signal this to a client other than
through the use of the truncate bit.
A malicious authoritative server could include a large number of
extra records (and associated DNSSEC information) and attempt to DoS
the recursive by making it do lots of DNSSEC validation. However,
this is not considered a realistic threat; CPU for validation is
cheap compared to bandwidth. This can be mitigated by allowing the
recursive resolver to ignore Additional records whenever it considers
itself under attack or its CPU resources are otherwise over-
committed.
This specification requires that the all of the Additional records
are signed, and all necessary DNSSEC information for validation be
included to avoid cache poisoning attacks.
11. Acknowledgements
The authors to thank Mark Andrews, John Dickinson, Kazunori Fujiwara,
Bob Harold, John Heidemann, and Tony Finch. The authors apologize in
advance for others who contributed, but who we managed to forget.
12. Normative References
[Ref.Bellovin]
Bellovin, S., "Using the Domain Name System for System
Break-Ins", 1995,
<https://www.cs.columbia.edu/~smb/papers/dnshack.pdf>.
[RFC1034] Mockapetris, P., "Domain names - concepts and facilities",
STD 13, RFC 1034, DOI 10.17487/RFC1034, November 1987,
<http://www.rfc-editor.org/info/rfc1034>.
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119, DOI 10.17487/
RFC2119, March 1997,
<http://www.rfc-editor.org/info/rfc2119>.
[RFC2181] Elz, R. and R. Bush, "Clarifications to the DNS
Specification", RFC 2181, DOI 10.17487/RFC2181, July 1997,
<http://www.rfc-editor.org/info/rfc2181>.
Kumari, et al. Expires January 4, 2018 [Page 7]
Internet-Draft DNS Extra Answers July 2017
[RFC5395] Eastlake 3rd, D., "Domain Name System (DNS) IANA
Considerations", RFC 5395, DOI 10.17487/RFC5395, November
2008, <http://www.rfc-editor.org/info/rfc5395>.
[RFC6555] Wing, D. and A. Yourtchenko, "Happy Eyeballs: Success with
Dual-Stack Hosts", RFC 6555, DOI 10.17487/RFC6555, April
2012, <http://www.rfc-editor.org/info/rfc6555>.
[RFC7873] Eastlake 3rd, D. and M. Andrews, "Domain Name System (DNS)
Cookies", RFC 7873, DOI 10.17487/RFC7873, May 2016,
<http://www.rfc-editor.org/info/rfc7873>.
Appendix A. Changes / Author Notes.
[RFC Editor: Please remove this section before publication ]
From -04 to -05:
o In the deadline rush, Warren forgot to add Tale. Fixed.
o Some more text fixups and clarifications.
From -03 to -04:
o Some additional text explaining how this differs from solutions
which include multiple queries (you don't know what to ask until
you have received some answers).
From -02 to -03:
o Sat down and rewrote and cleaned up large sections of text.
o Changed name of RR from Additional to EXTRA (the term "Additional"
is overloaded in general)
o Clarified that stub resolvers and other clients MAY use this
specification.
o Attempted to clarify that the individual RRs are added to the
response, not the EXTRA record itself. The EXTRA RR can be
queried directly.
From -00 to -01:
o Nothing change in the template.
Kumari, et al. Expires January 4, 2018 [Page 8]
Internet-Draft DNS Extra Answers July 2017
Authors' Addresses
Warren Kumari
Google
1600 Amphitheatre Parkway
Mountain View, CA 94043
US
Email: warren@kumari.net
Zhiwei Yan
CNNIC
No.4 South 4th Street, Zhongguancun
Beijing 100190
P. R. China
Email: yanzhiwei@cnnic.cn
Wes Hardaker
USC/ISI
P.O. Box 382
Davis, CA 95617
US
Email: ietf@hardakers.net
David C Lawrence
Akamai Technologies
150 Broadway
Cambridge, MA 02142-1054
US
Email: tale@akamai.com
Kumari, et al. Expires January 4, 2018 [Page 9]
