CURRENT MEETING REPORT
Minutes of the CIDR Deployment Working Group (CIDRD)
Reported by Geoff Huston, Telstra
1. Report on IP Address Space Usage and Lifetime (presented by Tony Li
and Frank Solensky)
1.a. Tony Li's Analysis
Tony summarized the IPv4 address space allocations,
indicating total address allocations over time.
The recent address allocation metrics were related to linear
forward extrapolation and the slight reduction in demand
levels across 1995 implied that this linear extrapolation
implied an increase in expectation of address space
lifetime.
At the San Jose IETF (Dec 94), Tony had estimated a
lifetime of the year 2008 +/- 3 years. By the Danvers IETF
(Apr 95), growth slope appeared to drop, which has been
constant at a slightly reduced linear tangent across 1995.
The projected lifetime is the year 2018 +/- 8 years.
The basis of these figures was discussed, noting that
Regional Registry allocations were reflected as single
anomalies in these figures, and further regional block
allocations may result in some degree of variance in this
projection.
The slides are to be found at
ftp://ftp.cisco.com/tli/ietf.slides.ps
1.b. Frank Solensky's Analysis
Frank presented his slides on similar projections. The
techniques for statistical extrapolation use a logistical
model where there is an implicit assumption of saturation
of demand at some point in the future. The extrapolation
attempts to predict the time of saturation of demand and
the level of resource use at the saturation point.
The analysis of the 128/2 space indicated that demand
saturation was already occuring, and that this would
stabilize at 62% of the 128/2 address space.
Analysis of the 192/3 space indicated that due to 1992-1995
data and 1994-1995 data, there were two possible
interpretations.
The full data set ("raw") yields a logistic interpretation of
saturation levels of 38% of the 192/3 address space
achieved by 1997 and a 95% confidence level of +- 2%.
The partial data set ("smoothed") yields a logistic
interpretation of a saturation level of 90% of the 192/3
address space achieved by 2006, with a 95% confidence
level of +- 35%.
These slides are reported (by Frank) to reside at
ftp://research.ftp.com/~solensky/cidrd.ps
However, the minute taker was unable to locate this
archive.
The logistic assumptions were challenged, in so far as the
observation was made that there was no natural limit to
demand other than exhaustion of the unallocated address
pool.
1.c 0/1 Space Report
Bill Manning noted that in the previous six months 13% of
the 0/1 space has been recovered through voluntary
returning of address space to the address registry.
2. Report on Routing Table Growth (Erik-Jan Bos)
Erik-Jan reported on the growth in the number of prefixes present in
routers. The number of total nets continues to grow, as is the number
of visible AS numbers in the routing tables.
The routing table was reported to contain 30,984 entries, an increase
of 10% across the previous four months.
62% of AS numbers are advertising CIDR routes, which is an
improvement since the Stockholm IETF (July 1995), and 77% of ASs
announcing more than one route are advertising CIDR routes.
An analysis of the routing entries indicate that the routing table is
holding entries of some 822,436,999 hosts and using current estimates
of Internet host population of some 8,000,000 connected hosts, the
total address space utilization efficiency is currently at 1%.
Since January 1994, Erik-Jan has been maintaining a database, with
entries for each hour, of the number of BGP entries in
Amsterdam1.dante.net. This plot indicates a recent reduction in
growth levels, although a strong linear trend is evident in the figure
during the period between June 1994-December 1995. This indicates
that the CIDR effort has to date managed to reduce the exponential
rate of routing table growth to a linear growth rate of 8,000 entries
per year over the past 18 months.
Erik-Jan�s slides are located at:
http://surver.wind.surfnet.nl/~bos/cd/cidrd.html
3. Documents
3.1 Appeal to return unused address space
This document has been hummed by the working group and will
be submitted to the IESG for publication as a BCP document.
3.2 RFC1597bis
This document has been hummed by the working group. The
issue of DNS configuration was raised where a DNS
environment was intended to be structured with both public
Internet and 1597 environment visibility. The document is to be
revised by Yakov Rekhter with the issue noted in a warning
paragraph and passed through the working group for final
review as a BCP submission.
3.3 Address Ownership
This document was reviewed in terms of security considerations
where renumbering may impact on the correct operation of
security environments. The document is to be edited with this
matter noted and then passed through the working group for
review.
3.4 Class A subnet deployment considerations
This document was hummed by the working group and is to be
passed to the IESG as a Working Group Informational RFC.
3.5 Net 39 experiment report
This document is being revised and Bill Manning will pass it to
the CIDRD Working Group for review within two months for
consideration as a Working Group Informational RFC.
3.6 CNAME extensions to in-addr DNS
This document is being revised to accomodate comments relating
to the operation of the DNS and will be tracked through the
DNSIND Working Group.
4. Non-Local Aggregation
The group discussed the use of Proxy Aggregation techniques in terms of
where this can be used within the routing space to reduce overall
routing table size. Tools for detection of aggregation conditions were
considered a possibility here and such routing table analysis tools will
be tested by working group members.
The Working Group will progress this through the preparation of an
internet draft which describes the cases where Proxy Aggregaton can be
undertaken without impact on the policy integrity of the routing space,
and the conditions where such indirect aggregation is not feasible.
5. Charging for Routing Advertisements
Yakov Rekhter presented the issues which are relevant to
consideration of charging for routing advertisements. The rationale
presented was that in order to improve the aggregation of information
within the routing space, to improve the overall efficiency of address
space utilization and improve the operational efficiency of the routing
space techncal solutions were potentially inadequate drivers, and that
introduction of an economic factor into this space would be a strong
positive feedback for achievement of these internet-wide objectives.
The presentation examined route charges as a component of bilateral
provider interaction and examined the cases where such interaction
would take place. This covers the push situation where routing
advertisements are pushed in the direction of a default-less routing
point, and the pull situation where the routing advertisements are
pulled outward from the default-less routing point.
