Internet Info 1997 December/Internet_Info_CD-ROM_Walnut_Creek_December_1997.iso/ietf/ale/ale-minutes-94mar.txt

Text File | 1994-05-13 | 8KB | 215 lines

CURRENT_MEETING_REPORT_

Reported by Erik-Jan Bos/SURFnet

Minutes of the Address Lifetime Expectations Working Group (ALE)


Agenda Bashing

Tony Li added one item to the agenda, IPng Timetable, at the request of
the area director.


Corrections to Models

During the ALE BOF at the Houston IETF, models were presented to look at
the expected lifetime of IPv4.  The intention of this agenda item is to
review the models currently used and make corrections if and when
needed.

The volume of the routing tables is still increasing.  Lines on the
graph show a million routes in several years time (Tony Li:  ``This is
scary'').  The question is, what will the effect of CIDR be?  Some
curves are trying to show the effect of CIDR, but it is only guessing
since no real data is available.  The current curves are based on data
beginning in September 1992.  The coming weeks and months will be
interesting starting points for new curves.

Most of today's CIDR announcements are blocks of class C addresses,
however, some blocks of Bs are already announced from Europe.

Curtis Villamizar did a dump of his routing tables and analyzed the
data.  Tony Bates and Curtis will be looking into the routing table
dump.  The dumps will be made on a weekly basis.  Regionals that are
willing to make their dumps available should provide pointers to the
appropriate FTP directory.  Other people are encouraged to dump their
routing tables and analyze the data.


Efficiency

Tony Bates reported that a look into the DNS was done for Europe to
determine the efficiency (percentage of usage) of the assigned address
space.  There are good reasons why this is not accurate (e.g.
firewalls, just not registering in DNS and other usage of IP like
NetBIOS add to the inaccuracy).  It is clear, however, that this is the
only source available.

Looking at improving efficiency is relatively easy for networks that are
new, but ``renumbering'' is not a lost battle.  Tools need to be
developed so that organizations can grow their networks without needing
more address space.  People should be given the means to be efficient.
Furthermore, there should be documents to help people.  Currently there
is no pointer to give to people.

Havard Eidnes volunteered to start this effort from a draft document he
wrote.  There is software that Charley Kline wrote that allowed him to
get approximately 30% efficiency in the assignment of host numbers at
the University of Illinois.  This should be made available at some
common location (e.g., Merit) after some restyling for portability.
Frank Solensky spoke with Charley a few minutes after the meeting.
Charley had just completed some of the finishing touches on his software
the previous week and was happy to make it available.


RFC 1597

Marten Terpstra reported on RFC 1597, an Informational RFC. Although
this is not a standards-track RFC, it is considered important because
the numbers are now available and the issue is being discussed.  It is
now possible to use address space inside a company, without needing a
review by another company to judge your efficiency.

The basic idea is simple:  assign network numbers that are non-unique,
but that are never going to be connected to the Internet.

The numbers are:


   o 10.0.0.0 - 10.255.255.255
   o 172.16.0.0 - 172.31.255.255
   o 192.168.0.0 - 192.168.255.255


Pros:


   o Save address space for the Internet
   o Use as much as you like
   o Perhaps some security advantage


Cons:


   o Connecting to the Internet or other private Internet needs
     renumbering

   o Traffic becomes untraceable when leaking

   o DNS and other things need more thinking

   o Generally something for an ``experienced'' network manager


If Network Address Translators (NAT boxes) are being used, this RFC
might have a large beneficial impact on the Internet.  This RFC is also
useful for sites that are definitely not going to connect.

The DNS consequences for deploying this RFC are not clear yet.  There
should be an additional RFC describing DNS consideration for RFC 1597.


IPng Timetable

The area director asked the working group to come up with a
``guesstimate of'' how long IPv4 will last.  Hand grenade tossing:


   o IP addresses will be available until 2008 plus or minus three
     years.  The assumptions should be clear, e.g., there are users out
     there that cannot be served right now.  If they are connecting this
     will become different.

   o Core routers are already running out of memory.  64Meg routers will
     buy another two years (plus or minus six months).  In an optimal
     CIDR world, there would probably be more time.


From this, no firm date can possibly be given, it all depends on the
success of CIDR and the cooperation of all service providers.

NAT boxes might also help CIDR, since a service provider is then able to
announce a non-CIDR-isable customer inside his own CIDR block.  This
will be studied further.


Action Items

Tony Bates and Curtis Villamizar  Conduct and analyze routing table
                                  dumps.

Havard Eidnes                     Brush up the document on improving
                                  efficiency.


Attendees

Juha-Pekka Ahopelto      [e-mail]
Mark Allyn               [e-mail]
Susie Armstrong          [e-mail]
Bashir Ashrafi           [e-mail]
Dennis Baker             [e-mail]
Fred Baker               [e-mail]
Tony Bates               [e-mail]
Jim Beers                [e-mail]
Nutan Behki              [e-mail]
Erik-Jan Bos             [e-mail]
Brad Burdick             [e-mail]
Joesph Burrescia         [e-mail]
Frank Cannata            [e-mail]
Greg Celmainis           [e-mail]
Brett Chappell           [e-mail]
Robert Christ            [e-mail]
Michael Collins          [e-mail]
Matt Crawford            [e-mail]
John Curran              [e-mail]
Sean Doran               [e-mail]
Kjeld Borch Egevang      [e-mail]
Havard Eidnes            [e-mail]
Nasser El-Aawar          [e-mail]
H. Tom Fitzpatrick       [e-mail]
Vince Fuller             [e-mail]
Steve Fulling            [e-mail]
Eugene Hastings          [e-mail]
Ian Heavens              [e-mail]
Jeff Hodges              [e-mail]
Matthew Jonson           [e-mail]
Sean Kennedy             [e-mail]
Edwin King               [e-mail]
So Young Lee             [e-mail]
Tony Li                  [e-mail]
Kim Long                 [e-mail]
Jun Matsukata            [e-mail]
Gerry Meyer              [e-mail]
Keith Mitchell           [e-mail]
Pushpendra Mohta         [e-mail]
Dennis Morris            [e-mail]
Rina Nathaniel           [e-mail]
Phil Nesser              [e-mail]
Ngoc-Lan Nguyen          [e-mail]
Michael O'Dell           [e-mail]
Andrew Partan            [e-mail]
Michael Patton           [e-mail]
Rex Pugh                 [e-mail]
Ron Roberts              [e-mail]
Robert Roden             [e-mail]
Duncan Rogerson          [e-mail]
Michal Rozenthal         [e-mail]
Steven Schnell           [e-mail]
Tim Seaver               [e-mail]
Frank Solensky           [e-mail]
John Tavs                [e-mail]
Marten Terpstra          [e-mail]
Jerry Toporek            [e-mail]
Paul Traina              [e-mail]
Wendell Turner           [e-mail]
Gary Veum                [e-mail]
Maria Vistoli            [e-mail]
Justin Walker            [e-mail]
Geoff White              [e-mail]
Jane Wojcik              [e-mail]