MoinQ: attachment:rfc4892.txt of form-example

   1 Network Working Group                                           S. Woolf
   2 Request for Comments: 4892             Internet Systems Consortium, Inc.
   3 Category: Informational                                        D. Conrad
   4                                                                    ICANN
   5                                                                June 2007
   6 
   7 
   8     Requirements for a Mechanism Identifying a Name Server Instance
   9 
  10 Status of This Memo
  11 
  12    This memo provides information for the Internet community.  It does
  13    not specify an Internet standard of any kind.  Distribution of this
  14    memo is unlimited.
  15 
  16 Copyright Notice
  17 
  18    Copyright (C) The IETF Trust (2007).
  19 
  20 Abstract
  21 
  22    With the increased use of DNS anycast, load balancing, and other
  23    mechanisms allowing more than one DNS name server to share a single
  24    IP address, it is sometimes difficult to tell which of a pool of name
  25    servers has answered a particular query.  A standardized mechanism to
  26    determine the identity of a name server responding to a particular
  27    query would be useful, particularly as a diagnostic aid for
  28    administrators.  Existing ad hoc mechanisms for addressing this need
  29    have some shortcomings, not the least of which is the lack of prior
  30    analysis of exactly how such a mechanism should be designed and
  31    deployed.  This document describes the existing convention used in
  32    some widely deployed implementations of the DNS protocol, including
  33    advantages and disadvantages, and discusses some attributes of an
  34    improved mechanism.
  35 
  36 1.  Introduction and Rationale
  37 
  38    Identifying which name server is responding to queries is often
  39    useful, particularly in attempting to diagnose name server
  40    difficulties.  This is most obviously useful for authoritative
  41    nameservers in the attempt to diagnose the source or prevalence of
  42    inaccurate data, but can also conceivably be useful for caching
  43    resolvers in similar and other situations.  Furthermore, the ability
  44    to identify which server is responding to a query has become more
  45    useful as DNS has become more critical to more Internet users, and as
  46    network and server deployment topologies have become more complex.
  47 
  48 
  49 
  50 
  51 
  52 Woolf & Conrad               Informational                      [Page 1]
  53 
  54 RFC 4892                        Serverid                       June 2007
  55 
  56 
  57    The conventional means for determining which of several possible
  58    servers is answering a query has traditionally been based on the use
  59    of the server's IP address as a unique identifier.  However, the
  60    modern Internet has seen the deployment of various load balancing,
  61    fault-tolerance, or attack-resistance schemes such as shared use of
  62    unicast IP addresses as documented in [RFC3258].  An unfortunate side
  63    effect of these schemes has been to make the use of IP addresses as
  64    identifiers associated with DNS (or any other) service somewhat
  65    problematic.  Specifically, multiple dedicated DNS queries may not go
  66    to the same server even though sent to the same IP address.  Non-DNS
  67    methods such as ICMP ping, TCP connections, or non-DNS UDP packets
  68    (such as those generated by tools like "traceroute"), etc., may well
  69    be even less certain to reach the same server as the one which
  70    receives the DNS queries.
  71 
  72    There is a well-known and frequently-used technique for determining
  73    an identity for a nameserver more specific than the possibly-non-
  74    unique "server that answered the query I sent to IP address A.B.C.D".
  75    The widespread use of the existing convention suggests a need for a
  76    documented, interoperable means of querying the identity of a
  77    nameserver that may be part of an anycast or load-balancing cluster.
  78    At the same time, however, it also has some drawbacks that argue
  79    against standardizing it as it's been practiced so far.
  80 
  81 2.  Existing Conventions
  82 
  83    For some time, the commonly deployed Berkeley Internet Name Domain
  84    (BIND) implementation of the DNS protocol suite from the Internet
  85    Systems Consortium [BIND] has supported a way of identifying a
  86    particular server via the use of a standards-compliant, if somewhat
  87    unusual, DNS query.  Specifically, a query to a recent BIND server
  88    for a TXT resource record in class 3 (CHAOS) for the domain name
  89    "HOSTNAME.BIND." will return a string that can be configured by the
  90    name server administrator to provide a unique identifier for the
  91    responding server.  (The value defaults to the result of a
  92    gethostname() call).  This mechanism, which is an extension of the
  93    BIND convention of using CHAOS class TXT RR queries to sub-domains of
  94    the "BIND." domain for version information, has been copied by
  95    several name server vendors.
  96 
  97    A refinement to the BIND-based mechanism, which dropped the
  98    implementation-specific label, replaces "BIND." with "SERVER.".  Thus
  99    the query label to learn the unique name of a server may appear as
 100    "ID.SERVER.".
 101 
 102    (For reference, the other well-known name used by recent versions of
 103    BIND within the CHAOS class "BIND." domain is "VERSION.BIND.".  A
 104    query for a CHAOS TXT RR for this name will return an
 105 
 106 
 107 
 108 Woolf & Conrad               Informational                      [Page 2]
 109 
 110 RFC 4892                        Serverid                       June 2007
 111 
 112 
 113    administratively defined string which defaults to the software
 114    version of the server responding.  This is, however, not generally
 115    implemented by other vendors.)
 116 
 117 2.1.  Advantages
 118 
 119    There are several valuable attributes to this mechanism, which
 120    account for its usefulness.
 121 
 122    1.  The "HOSTNAME.BIND." or "ID.SERVER." query response mechanism is
 123        within the DNS protocol itself.  An identification mechanism that
 124        relies on the DNS protocol is more likely to be successful
 125        (although not guaranteed) in going to the same system as a
 126        "normal" DNS query.
 127 
 128    2.  Since the identity information is requested and returned within
 129        the DNS protocol, it doesn't require allowing any other query
 130        mechanism to the server, such as holes in firewalls for
 131        otherwise-unallowed ICMP Echo requests.  Thus it is likely to
 132        reach the same server over a path subject to the same routing,
 133        resource, and security policy as the query, without any special
 134        exceptions to site security policy.
 135 
 136    3.  It is simple to configure.  An administrator can easily turn on
 137        this feature and control the results of the relevant query.
 138 
 139    4.  It allows the administrator complete control of what information
 140        is given out in the response, minimizing passive leakage of
 141        implementation or configuration details.  Such details are often
 142        considered sensitive by infrastructure operators.
 143 
 144 2.2.  Disadvantages
 145 
 146    At the same time, there are some serious drawbacks to the CHAOS/TXT
 147    query mechanism that argue against standardizing it as it currently
 148    operates.
 149 
 150    1.  It requires an additional query to correlate between the answer
 151        to a DNS query under normal conditions and the supposed identity
 152        of the server receiving the query.  There are a number of
 153        situations in which this simply isn't reliable.
 154 
 155    2.  It reserves an entire class in the DNS (CHAOS) for what amounts
 156        to one zone.  While CHAOS class is defined in [RFC1034] and
 157        [RFC1035], it's not clear that supporting it solely for this
 158        purpose is a good use of the namespace or of implementation
 159        effort.
 160 
 161 
 162 
 163 
 164 Woolf & Conrad               Informational                      [Page 3]
 165 
 166 RFC 4892                        Serverid                       June 2007
 167 
 168 
 169    3.  The initial and still common form, using "BIND.", is
 170        implementation specific.  BIND is one DNS implementation.  At the
 171        time of this writing, it is probably most prevalent for
 172        authoritative servers.  This does not justify standardizing on
 173        its ad hoc solution to a problem shared across many operators and
 174        implementors.  Meanwhile, the aforementioned refinement changes
 175        the query label but preserves the ad hoc CHAOS/TXT mechanism.
 176 
 177    4.  There is no convention or shared understanding of what
 178        information an answer to such a query for a server identity could
 179        or should contain, including a possible encoding or
 180        authentication mechanism.
 181 
 182    5.  Hypothetically, since DNSSEC has been defined to cover all DNS
 183        classes, the TXT RRs returned in response to the "ID.SERVER."
 184        query could be signed, which has the advantages described in
 185        [RFC4033].  However, since DNSSEC deployment for the CHAOS class
 186        is neither existent nor foreseeable, and since the "ID.SERVER."
 187        TXT RR is expected to be unique per server, this would be
 188        impossible in practice.
 189 
 190    The first of the listed disadvantages may be technically the most
 191    serious.  It argues for an attempt to design a good answer to the
 192    problem, "I need to know what nameserver is answering my queries",
 193    not simply a convenient one.
 194 
 195 3.  Characteristics of an Implementation Neutral Convention
 196 
 197    The discussion above of advantages and disadvantages to the
 198    "HOSTNAME.BIND." mechanism suggest some requirements for a better
 199    solution to the server identification problem.  These are summarized
 200    here as guidelines for any effort to provide appropriate protocol
 201    extensions:
 202 
 203    1.  The mechanism adopted must be in-band for the DNS protocol.  That
 204        is, it needs to allow the query for the server's identifying
 205        information to be part of a normal, operational query.  It should
 206        also permit a separate, dedicated query for the server's
 207        identifying information.  But it should preserve the ability of
 208        the CHAOS/TXT query-based mechanism to work through firewalls and
 209        in other situations where only DNS can be relied upon to reach
 210        the server of interest.
 211 
 212    2.  The new mechanism should not require dedicated namespaces or
 213        other reserved values outside of the existing protocol mechanisms
 214        for these, i.e., the OPT pseudo-RR.  In particular, it should not
 215        propagate the existing drawback of requiring support for a CLASS
 216 
 217 
 218 
 219 
 220 Woolf & Conrad               Informational                      [Page 4]
 221 
 222 RFC 4892                        Serverid                       June 2007
 223 
 224 
 225        and top level domain in the authoritative server (or the querying
 226        tool) to be useful.
 227 
 228    3.  Support for the identification functionality should be easy to
 229        implement and easy to enable.  It must be easy to disable and
 230        should lend itself to access controls on who can query for it.
 231 
 232    4.  It should be possible to return a unique identifier for a server
 233        without requiring the exposure of information that may be non-
 234        public and considered sensitive by the operator, such as a
 235        hostname or unicast IP address maintained for administrative
 236        purposes.
 237 
 238    5.  It should be possible to authenticate the received data by some
 239        mechanism analogous to those provided by DNSSEC.  In this
 240        context, the need could be met by including encryption options in
 241        the specification of a new mechanism.
 242 
 243    6.  The identification mechanism should not be implementation-
 244        specific.
 245 
 246 4.  IANA Considerations
 247 
 248    This document proposes no specific IANA action.  Protocol extensions,
 249    if any, to meet the requirements described are out of scope for this
 250    document.  A proposed extension, specified and adopted by normal IETF
 251    process, is described in [NSID], including relevant IANA action.
 252 
 253 5.  Security Considerations
 254 
 255    Providing identifying information as to which server is responding to
 256    a particular query from a particular location in the Internet can be
 257    seen as information leakage and thus a security risk.  This motivates
 258    the suggestion above that a new mechanism for server identification
 259    allow the administrator to disable the functionality altogether or
 260    partially restrict availability of the data.  It also suggests that
 261    the server identification data should not be readily correlated with
 262    a hostname or unicast IP address that may be considered private to
 263    the nameserver operator's management infrastructure.
 264 
 265    Propagation of protocol or service meta-data can sometimes expose the
 266    application to denial of service or other attack.  As the DNS is a
 267    critically important infrastructure service for the production
 268    Internet, extra care needs to be taken against this risk for
 269    designers, implementors, and operators of a new mechanism for server
 270    identification.
 271 
 272 
 273 
 274 
 275 
 276 Woolf & Conrad               Informational                      [Page 5]
 277 
 278 RFC 4892                        Serverid                       June 2007
 279 
 280 
 281    Both authentication and confidentiality of server identification data
 282    are potentially of interest to administrators -- that is, operators
 283    may wish to make such data available and reliable to themselves and
 284    their chosen associates only.  This constraint would imply both an
 285    ability to authenticate it to themselves and to keep it private from
 286    arbitrary other parties, which leads to characteristics 4 and 5 of an
 287    improved solution.
 288 
 289 6.  Acknowledgements
 290 
 291    The technique for host identification documented here was initially
 292    implemented by Paul Vixie of the Internet Software Consortium in the
 293    Berkeley Internet Name Daemon package.  Comments and questions on
 294    earlier versions were provided by Bob Halley, Brian Wellington,
 295    Andreas Gustafsson, Ted Hardie, Chris Yarnell, Randy Bush, and
 296    members of the ICANN Root Server System Advisory Committee.  The
 297    newest version takes a significantly different direction from
 298    previous versions, owing to discussion among contributors to the
 299    DNSOP working group and others, particularly Olafur Gudmundsson, Ed
 300    Lewis, Bill Manning, Sam Weiler, and Rob Austein.
 301 
 302 7.  References
 303 
 304 7.1.  Normative References
 305 
 306    [RFC1034]  Mockapetris, P., "Domain Names - Concepts and Facilities",
 307               STD 13, RFC 1034, November 1987.
 308 
 309    [RFC1035]  Mockapetris, P., "Domain Names - Implementation and
 310               Specification", STD 13, RFC 1035, November 1987.
 311 
 312    [RFC3258]  Hardie, T., "Distributing Authoritative Name Servers via
 313               Shared Unicast Addresses", RFC 3258, April 2002.
 314 
 315 7.2.  Informative References
 316 
 317    [BIND]     ISC, "BIND 9 Configuration Reference".
 318 
 319    [NSID]     Austein, R., "DNS Name Server Identifier Option (NSID)",
 320               Work in Progress, June 2006.
 321 
 322    [RFC4033]  Arends, R., Austein, R., Larson, M., Massey, D., and S.
 323               Rose, "DNS Security Introduction and Requirements", RFC
 324               4033, March 2005.
 325 
 326 
 327 
 328 
 329 
 330 
 331 
 332 Woolf & Conrad               Informational                      [Page 6]
 333 
 334 RFC 4892                        Serverid                       June 2007
 335 
 336 
 337 Authors' Addresses
 338 
 339    Suzanne Woolf
 340    Internet Systems Consortium, Inc.
 341    950 Charter Street
 342    Redwood City, CA  94063
 343    US
 344 
 345    Phone: +1 650 423-1333
 346    EMail: woolf@isc.org
 347    URI:   http://www.isc.org/
 348 
 349 
 350    David Conrad
 351    ICANN
 352    4676 Admiralty Way
 353    Marina del Rey, CA  90292
 354    US
 355 
 356    Phone: +1 310 823 9358
 357    EMail: david.conrad@icann.org
 358    URI:   http://www.iana.org/
 359 
 360 
 361 
 362 
 363 
 364 
 365 
 366 
 367 
 368 
 369 
 370 
 371 
 372 
 373 
 374 
 375 
 376 
 377 
 378 
 379 
 380 
 381 
 382 
 383 
 384 
 385 
 386 
 387 
 388 Woolf & Conrad               Informational                      [Page 7]
 389 
 390 RFC 4892                        Serverid                       June 2007
 391 
 392 
 393 Full Copyright Statement
 394 
 395    Copyright (C) The IETF Trust (2007).
 396 
 397    This document is subject to the rights, licenses and restrictions
 398    contained in BCP 78, and except as set forth therein, the authors
 399    retain all their rights.
 400 
 401    This document and the information contained herein are provided on an
 402    "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS
 403    OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND
 404    THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS
 405    OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF
 406    THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED
 407    WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
 408 
 409 Intellectual Property
 410 
 411    The IETF takes no position regarding the validity or scope of any
 412    Intellectual Property Rights or other rights that might be claimed to
 413    pertain to the implementation or use of the technology described in
 414    this document or the extent to which any license under such rights
 415    might or might not be available; nor does it represent that it has
 416    made any independent effort to identify any such rights.  Information
 417    on the procedures with respect to rights in RFC documents can be
 418    found in BCP 78 and BCP 79.
 419 
 420    Copies of IPR disclosures made to the IETF Secretariat and any
 421    assurances of licenses to be made available, or the result of an
 422    attempt made to obtain a general license or permission for the use of
 423    such proprietary rights by implementers or users of this
 424    specification can be obtained from the IETF on-line IPR repository at
 425    http://www.ietf.org/ipr.
 426 
 427    The IETF invites any interested party to bring to its attention any
 428    copyrights, patents or patent applications, or other proprietary
 429    rights that may cover technology that may be required to implement
 430    this standard.  Please address the information to the IETF at
 431    ietf-ipr@ietf.org.
 432 
 433 Acknowledgement
 434 
 435    Funding for the RFC Editor function is currently provided by the
 436    Internet Society.
 437 
 438 
 439 
 440 
 441 
 442 
 443 
 444 Woolf & Conrad               Informational                      [Page 8]
 445