[Docs] [txt|pdf] [Tracker] [Email] [Diff1] [Diff2] [Nits]

Versions: 00 01 02

DNS Operations(dnsop)                                        K. Fujiwara
Internet-Draft                                                      JPRS
Intended status: Informational                          January 20, 2014
Expires: July 24, 2014

            Side effect of DNSSEC: an increase of DS queries


   An increase of periodic DS queries is observed at top level domain
   (TLD) DNS servers.  The reason of the increase is low NCACHE TTL
   value and DS nonexistence.  This memo presents issues with DNSSEC and
   small NCACHE TTL value, including possible countermeasures in order
   to prepare future increase of DS queries.

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 July 24, 2014.

Copyright Notice

   Copyright (c) 2014 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.

Fujiwara                  Expires July 24, 2014                 [Page 1]

Internet-Draft           Increase of DS queries             January 2014

Table of Contents

   1.  Introduction  . . . . . . . . . . . . . . . . . . . . . . . .   2
   2.  Problem statement . . . . . . . . . . . . . . . . . . . . . .   2
   3.  Possible affected domain names  . . . . . . . . . . . . . . .   3
   4.  Possible measures . . . . . . . . . . . . . . . . . . . . . .   3
     4.1.  Dummy DS idea . . . . . . . . . . . . . . . . . . . . . .   4
   5.  References  . . . . . . . . . . . . . . . . . . . . . . . . .   5

1.  Introduction

   A significant increase of DS queries is observed at JP TLD DNS
   servers.  4.5% of queries are DS queries at JP TLD DNS servers in
   Dec., 2013 and they are still increasing.  Almost all query names of
   DS queries are unsigned zone cuts.  These DS queries are useless for
   DNSSEC validation because they are unsigned delegations.  Very small
   number of IP addresses send most of DS queries and the DS queries are
   periodic.  The reason of the increase is low NCACHE TTL value and DS
   nonexistence.  Details are described in Section 2.  Possible affected
   domain names are described in Section 3.  Possible countermeasures
   are described in Section 4.

2.  Problem statement

   Many TLDs have supported DNSSEC.  However, many delegations do not
   have DS resource records.  Some of full-resolvers support DNSSEC

   The conditions of the DS query increase are as follows.

   o  TLD's TTL value is relatively high, e.g., 86400.

   o  TLD's NCACHE TTL value is low, e.g., 900.

   o  There are many popular query names whose resource record TTLs are
      low, e.g., 300, and they are unsigned.

   o  DNSSEC validators receive queries of popular names frequently,
      e.g. every 5 minutes.

   An unsigned delegation does not have a DS RR in its TLD zone.  DNSSEC
   validation process starts when the validator receives a query and it
   does not exist in the validator's cache.  DNSSEC validators need to
   know DS RR existence for each query name.  The DS RR nonexistence
   information is cached within NCACHE TTL.  As a result, each DNSSEC
   validator may send DS queries to TLD DNS servers one zone cut per
   NCACHE TTL seconds.

Fujiwara                  Expires July 24, 2014                 [Page 2]

Internet-Draft           Increase of DS queries             January 2014

   This phenomena is DNSSEC protocol and DNS parameter issue.  DS
   queries will increase as DNSSEC validators will increase.

   JP TLD case, NS and glue TTL is 86400 and NCACHE TTL is 900.  There
   are many popular names which are unsigned domain names and whose TTLs
   are low.  TTL of "www.yahoo.co.jp" A is 60 (CNAME TTL is 900 and TTL
   of aliased name is 60) and TTL of "www.google.co.jp" A is 300.  Busy
   full-resolvers receive both queries every minutes or more.  When a
   busy full-resolver enables DNSSEC validation, it will send
   "yahoo.co.jp" and "google.co.jp" DS queries every 900 seconds.
   "yahoo.co.jp" NS and "google.co.jp" NS are cached in a day (86400
   seconds).  As a result, queries to JP DNS servers may increase 96
   (86400 / 900) times at the maximum.  This is DNSSEC protocol and
   parameter issue.

3.  Possible affected domain names

   Possible affected domain names are delegation centric domain names
   which support DNSSEC, whose NCACHE TTL is low, and which has popular
   domain names which are not signed and use low TTL values.

   TLDs:  com, net, org, jp use 900 as NCACHE TTL value.  Magnification
      is 96 or more.

   Reverse DNS:  193.in-addr.arpa uses 3600 as NCACHE TTL value.
      Magnification is 48.

   The root is affected a little because popular TLDs have already been
   signed and the magnification is not high, 8 or 24 (86400 / 10800 or
   86400 / 3600).

4.  Possible measures

   There are no good solutions and five possible measures to the

   1.  Reinforce DNS infrastructures.

   2.  Sign popular domain names.  If popular domain names are signed,
       their DS RRs are cached.  However, a TLD can not control them.
       Some TLDs have been trying to increase signed delegations by
       price or security campaigns.

   3.  Lengthen resource record TTL of popular names.  However, a TLD
       can not control.

   4.  Lengthen NCACHE TTL value.  However, the value is chosen by the
       TLD's policy and this approach can not stop the increase of DS

Fujiwara                  Expires July 24, 2014                 [Page 3]

Internet-Draft           Increase of DS queries             January 2014

       queries.  Section 5 of DNS NCACHE [RFC2308] recommends negative
       cache time limit as values of one to three hours.  Lengthening
       NCACHE TTL value over 10800 is useless.  Magnification can only
       be lowered.  (JP case, from 96 to 8 or 24.)

   5.  Update DNS/DNSSEC protocol to reduce unnecessary DS queries.
       There are some idea.

       1.  Changing validator's caching algorithms.

       2.  Adding dummy DS to popular unsigned delegations.  Details are
           described in Section 4.1.

   Without protocol modifications, we need to reinforce DNS
   infrastructures and try to increase signed delegations.

4.1.  Dummy DS idea

   "Adding dummy DS to popular unsigned delegations."  Dummy DS RR may
   be ignored by traditional DNSSEC validators and it indicates that the
   delegation is an unsigned delegation.  Dummy DS TTL value is
   controllable.  This proposal requires new digest type.

   Dummy DS RR will be ignored by traditional DNSSEC validators because
   Section 5.2 of DNSSEC Protocol [RFC4035] defines that the resolver
   should treat unknown digest type as no DS RRset exists.  BIND 9 and
   Unbound validators ignored dummy DS RR whose digest type is 255.

   However, there are many considerations.

   o  Dummy DS RRs may be treated as a DNSSEC error.  Google public DNS
      reports validation error at dummy DSs.  BIND 9 and Unbound
      validators ignore dummy DSs.  DNSSEC Protocol [RFC4035] may be

   o  Dummy DS RRs increase signing costs because most of TLDs use opt-
      out technique defined in NSEC3 [RFC5155] to reduce signed domain

   o  Newly added DS RRs may be used within dummy DSs' TTL seconds (for
      example, it will be 1 day).  Without dummy DS RRs, newly added DS
      RRs are used within NCACHE TTL (900 or 10800 seconds).

   o  Is it allowed that TLDs add dummy DS RRs without registrants'
      consent?  If adding dummy DS is same as 'NO DS', it is possible.
      Otherwise, TLDs cannot add dummy DS RRs without registrants'

Fujiwara                  Expires July 24, 2014                 [Page 4]

Internet-Draft           Increase of DS queries             January 2014

5.  References

   [RFC2308]  Andrews, M., "Negative Caching of DNS Queries (DNS
              NCACHE)", RFC 2308, March 1998.

   [RFC4035]  Arends, R., Austein, R., Larson, M., Massey, D., and S.
              Rose, "Protocol Modifications for the DNS Security
              Extensions", RFC 4035, March 2005.

   [RFC5155]  Laurie, B., Sisson, G., Arends, R., and D. Blacka, "DNS
              Security (DNSSEC) Hashed Authenticated Denial of
              Existence", RFC 5155, March 2008.

Author's Address

   Kazunori Fujiwara
   Japan Registry Services Co., Ltd.
   Chiyoda First Bldg. East 13F, 3-8-1 Nishi-Kanda
   Chiyoda-ku, Tokyo  101-0065

   Phone: +81 3 5215 8451
   EMail: fujiwara@jprs.co.jp

Fujiwara                  Expires July 24, 2014                 [Page 5]

Html markup produced by rfcmarkup 1.129d, available from https://tools.ietf.org/tools/rfcmarkup/