Re: ICANN seeks public comments on Fast Flux

[R Atkinson <ran.atkinson@xxxxxxxxx>]

While I am very happy to see this initial report, the initial
report on "Fast Flux DNS" fails to note clearly that a range
of very legitimate uses for DNS records with very low TTL
values exist.  For example, short TTL values for the DNS
A/PTR (or equivalent IPv6 DNS records) are quite useful
to provide mobility support.   If a DNS server is itself
mobile (e.g. aboard a ship, airplane, or vehicle), then
that DNS server's own A record might well need to have
a very short TTL value (e.g. minutes).  If one's network
is being renumbered, or is likely to be renumbered in the
near future, those are separate (and very legitimate) reasons
why A/PTR DNS records, or even MX/KX/other DNS records,
might have very short TTL values.


It is less obvious that DNS nameserver delegation records
(i.e. NS records) would need to have shorter TTL values
in ordinary daily operation.  This is a critical distinction
from the various examples provided above.  This distinction
is not crisply made in the current version of the ICANN paper.


The report should be edited to clarify the existence of
legitimate reasons to have DNS records with low TTL values,
including the sundry examples above, and to distinguish
clearly between cases where the a particular DNS record type
has a low TTL value for no obvious reason (e.g. possibly
the case of a NS record with a very low TTL value) and
where legitimate reasons might exist for other DNS
records to have a low TTL value (e.g. mobile hosts,
mobile DNS servers, mobile networks, ad-hoc networks
all might benefit from low TTL values for A/PTR,
IPv6-equivalent DNS record types, and MX/KX/SRV/
other non-NS record types).


There are multiple peer-reviewed papers on the use of
DNS as part of an Internet mobility solution.  To give
only a few examples (more papers on this exist; I am
an author of the 2nd & 3rd papers below):

"Mobile Host Location Tracking through DNS" by A. Pappas,
S. Hailes, et alia, Proceedings of IEEE London Communications
Symposium, August 2002.
  <http://www.ee.ucl.ac.uk/lcs/papers2002/LCS072.pdf>

"Mobility as an Integrated Service through the use of Naming",
by R. Atkinson, S. Bhatti, & S. Hailes, Proceedings of ACM
MobiArch Workshop, August 2007.
  <http://www.sigcomm.org/sigcomm2007/mobiarch/Atkinson_Naming.pdf>

"Mobility through Naming - Impact on DNS"
by R. Atkinson, S. Bhatti, & S. Hailes, Proceedings of ACM
MobiArch Workshop, August 2008.   <http://conferences.sigcomm.org/sigcomm/2008/workshops/mobiarch/papers/p7.pdf 
>
"Integrated DNS and IPv6 Mobility for Next-Generation Internet",
by W.K.G. Seah & S.K Tan, Institute for Infocomm Research,
Singapore.
  <http://www1.i2r.a-star.edu.sg/~winston/papers/IWNT2003-Integrated%20DNS%20and%20IPv6%20mobility.pdf 
>
"Comparative Performance Analysis of Domain Name based
Location Management" by A.S. Reaz & M. Atiquzzaman,
Proceedings of 2nd International Conference on Wireless
& Mobile Communications, IEEE, 2006.


Separately, I should note there is active research in this
area by several different groups.  Internet Mobility is a
very active research area, with a range of groups working
in the area and a wide range of conferences and journals.
Again, giving only a couple of examples of current research
projects in this area that are examining the use of DNS in
a mobility context:

"Identifier/Locator Network Protocol (ILNP)" project,
which I happen to be involved with; the project web site
is hosted by U. St Andrews:
   <http://ilnp.cs.st-andrews.ac.uk>

"Network Mobility" project at Lancaster University:
   <http://www.network-mobility.org/index.php?page=dns>
   <http://www.network-mobility.org/index.php?page=ippaging>

European Union's "IST Enable" project:
   <http://www.ist-enable.org>

MIT's "Migrate Internet Mobility project:
   <http://nms.lcs.mit.edu/projects/migrate>


There are also current IETF drafts relating to the use
of DNS as part of network/host mobility solutions,
for example:

"Locating IEEE 802.21 Mobility Servers using DNS",
by G. Bajko, Internet-Draft, 17 October 2008
  <http://tools.ietf.org/html/draft-ietf-mipshop-mos-dns-discovery-04>


So the initial paper on the "Fast Flux DNS" topic really needs
additional work to more crisply differentiate the various
legitimate uses for DNS records with short TTL values (several
of which are outlined in this email) from the other unfortunate
uses for DNS records with short TTL values.


Finally, it is not obvious (perhaps I overlooked something ?)
that the initial paper has been fully reviewed by the IETF's
DNS Extensions Working Group and also by other IETF WGs
(particularly mobility-related IETF WGs).  Such a review
by those IETF groups is important to ensure that not only
current DNS-related specifications and deployments,
but also emerging and anticipated DNS-related specifications
and deployments, are fully taken into account in the report.

Yours,

RJ Atkinson

DISCLAIMER: I never speak for my employer; the above are my own views.