Transcript IPv4 Address Lifetime Expectancy Revisited Geoff Huston September 2003 Presentation to the RIPE 46 Plenary Research activity supported by APNIC The Regional Internet Registries s do not.

IPv4 Address Lifetime Expectancy
Revisited
Geoff Huston
September 2003
Presentation to the RIPE 46 Plenary
Research activity
supported by APNIC
The Regional Internet Registries s do not make forecasts or
predictions about number resource lifetimes. The RIRs provide
statistics of what has been allocated. The following presentation
is a personal contribution based on extrapolation of RIR
allocation data.
IPv4 Address Lifetime
Expectancy



This was an IETF activity starting as part of
the Routing and Addressing (ROAD) activity
in the early 1990’s
The objective was to understand the rate of
allocation of IPv4 addresses and make some
predictions as to the date of eventual
exhaustion of the unallocated address pool
This is a re-visiting of this activity with
consideration of additional data derived from
the characteristics of the BGP routing table
The IPv4 Address Space



A 32 bit field spanning some 4.4B entries
The IETF, through standards actions, has
determined some space to be used for global
unicast, some for multicast and some held in
reserve
IANA has allocated some unicast space to the
RIRs for further allocation and assignment,
assigned some space directly, and reserved
some space for particular purposes
The IPv4 Top Level
Structure
Reserved 7.5% 19 /8
Multicast 6.2% 16 /8
Unicast 86.3%
221 /8
Modeling the Process

A number of views can be used to make
forward projections:



The rate at which IPv4 number blocks are
passed from IANA to the RIRs
The rate at which RIRs undertake
assignments of IPv4 address blocks to
LIRs and end users
The growth of the number of announced
addresses in the BGP routing table
Data Sets

IANA IPv4 Address Registry


RIR Stats files


Allocation of /8 blocks to RIRs and others
Allocation of blocks to LIRs
BGP Routing table

Amount of address space advertised as
reachable
IANA Allocations


The IPv4 address registry records the date
of each /8 allocation undertaken by the
IANA
This data has some inconsistencies, but is
stable enough to allow some form of
projection
IANA Registry Comments

The allocation dates for those address blocks
prior to 1995 are inaccurate




The earliest date is 1991, and a large block has
been recorded as allocated in 1993.
This is inconsistent with dates recorded in the RIR
stats files, which record allocations back to 1983
It would appear that there was a revision of the
IANA registry in the period 1991 – 1993, and the
IANA recorded dates are the revision dates
Useable dates appear to start from allocations
from 1995 onwards
IANA – Current Status
IETF Reserved 7.5%
Multicast 6.2%
Unicast - Allocated 51.1%
Unicast IANA Reserved 35.2%
IANA Allocations
IANA Allocations (/8)
200
150
100
50
0
Jan-91
Jan-92
Jan-93
Jan-94
Jan-95
Jan-96
Jan-97
Jan-98
Jan-99
Jan-00
Jan-01
Jan-02
Jan-03
IANA Projections
IANA Allocation Projection
200
150
100
50
0
Jan-91
Jan-93
Jan-95
Jan-97
Jan-99
Jan-01
Jan-03
Jan-05
Jan-07
Jan-09
Jan-11
Jan-13
Jan-15
Jan-17
Jan-19
IANA Projections

This projection of 2019 for IANA
address pool exhaustion is very
uncertain because of:



Sensitivity of allocation rate to prevailing
RIR assignment policies
Takeup of applications that require end-toend IP addressing vs use of NATs
Potential use of a further 16 /8s currently
reserved by the IETF
RIR Allocations

The RIR stats files records the date of
each allocation to an LIR, together with
the allocation details
RIR Allocations – Current
Status
Multicast, 16, 6%
IETF Reserved, 19, 7%
Unicast RIR - Allocated, 115.87,
46%
Unicast IANA Reserved, 90,
35%
Unicast RIR/IANA - Held, 15.13,
6%
RIR Allocations /8
Address Allocation Status - by /8
16000000
14000000
12000000
10000000
Reserved
IANA
Unallocated
Allocated
8000000
6000000
4000000
2000000
10
4
11
2
12
0
12
8
13
6
14
4
15
2
16
0
16
8
17
6
18
4
19
2
20
0
20
8
21
6
22
4
23
2
24
0
24
8
96
88
80
72
64
56
48
40
32
24
16
8
0
0
n03
Ja
n02
Ja
n01
Ja
n00
Ja
n99
Ja
n98
Ja
n97
Ja
n96
Ja
n95
Ja
n94
Ja
n93
Ja
n92
Ja
n91
Ja
n90
Ja
n89
Ja
n88
Ja
n87
Ja
n86
Ja
n85
Ja
n84
Ja
n83
Ja
RIR Allocations
RIR Allocations
200
150
100
50
0
Jan-29
Jan-27
Jan-25
Jan-23
Jan-21
Jan-19
Jan-17
Jan-15
Jan-13
Jan-11
Jan-09
Jan-07
Jan-05
Jan-03
Jan-01
Jan-99
Jan-97
Jan-95
Jan-93
Jan-91
Jan-89
Jan-87
Jan-85
Jan-83
RIR Projections
RIR Allocations - Projection
221
200
150
100
50
0
RIR Projections

This projections of 2026 for 221 /8s and 2029
for 237/8s has the same levels of uncertainty
as noted for the IANA projections
BGP Routing Table


The BGP routing table spans a set of
advertised addresses
A similar analysis of usage and
projection can be undertaken on this
date
The Route Views view
The AS1221 view
BGP Routing Table - Status
Multicast, 16, 6%
IETF Reserved, 19, 7%
Unicast BGP - Announced,
73.27, 29%
Unicast IANA Reserved, 90,
35%
Unicast RIR - Allocated, 42.6,
17%
Unicast RIR/IANA - Held, 15.13,
6%
BGP Address Allocations /8
Address Allocation Status - by /8
16000000
14000000
12000000
10000000
Reserved
IANA
Unallocated
Unadvertised
Advertised
8000000
6000000
4000000
2000000
96
10
4
11
2
12
0
12
8
13
6
14
4
15
2
16
0
16
8
17
6
18
4
19
2
20
0
20
8
21
6
22
4
23
2
24
0
24
8
88
80
72
64
56
48
40
32
24
16
8
0
0
Age of Unannounced Blocks
Age Distribution of Unannounced Address Space (/8s)
2.5
2
1.5
1
0.5
0
Jan-83
Jan-85
Jan-87
Jan-89
Jan-91
Jan-93
Jan-95
Jan-97
Jan-99
Jan-01
Jan-03
Age of Unannounced Blocks
(cumulative)
Cumulative Age Distribution of Unannounced Address Space (/8s)
35
30
25
20
15
10
5
0
Jan-83
Jan-85
Jan-87
Jan-89
Jan-91
Jan-93
Jan-95
Jan-97
Jan-99
Jan-01
Jan-03
May/03
Mar/03
Jan/03
Nov/02
Sep/02
Jul/02
May/02
Mar/02
Jan/02
Nov/01
Sep/01
Jul/01
May/01
Mar/01
Jan/01
Nov/00
Sep/00
Jul/00
May/00
Mar/00
Jan/00
Nov/99
BGP Address Span
BGP Table - Address Span
200
150
100
50
0
Jan/28
Jan/27
Jan/26
Jan/25
Jan/24
Jan/23
Jan/22
Jan/21
Jan/20
Jan/19
Jan/18
Jan/17
Jan/16
Jan/15
Jan/14
Jan/13
Jan/12
Jan/11
Jan/10
Jan/09
Jan/08
Jan/07
Jan/06
Jan/05
Jan/04
Jan/03
Jan/02
Jan/01
Jan/00
BGP Projections
BGP Announced Address Space - Projection
200
150
100
50
0
BGP Projections

This projection of 2027 (221 /8s) and
2028 (237 /8s) uses a 3 year baseline


This is much shorter baseline than the
IANA and RIR projections
There are, again, considerable
uncertainties associated with this
projection
Another look at that BGP
data:…


Comments received about this
projection have prompted me to review
the BGP address data and see if a more
detailed analysis of the BGP data
modifies this model
It appears to be the case that there is a
different view that can be formed from
the data:……
Another look at that BGP
data:…

Firstly, here’s the raw data – hourly
measurements over 3 years.
Another look at that BGP
data:…
BGP Data
1230000000
1180000000
1130000000
1080000000
1030000000
980000000
Nov-99
Feb-00
May-00
Aug-00
Nov-00
Feb-01
May-01
Aug-01
Nov-01
Feb-02
May-02
Aug-02
Nov-02
Feb-03
May-03
Aug-03
Another look at that BGP
data:…



The most obvious noise comes from
flaps in /8 advertisements.
The first step was to remove this noise
by recalculating the address data using
a fixed number of /8 advertisements
The value of 19 was used to select one
of the ‘tracks’ in the data
Another look at that BGP
data:…
Filter to 19 /8s
1230000000
1180000000
1130000000
1080000000
1030000000
980000000
Nov-99
Mar-00
Jul-00
Nov-00
Mar-01
Jul-01
Nov-01
Mar-02
Jul-02
Nov-02
Mar-03
Jul-03
Another look at that BGP
data:…


This is still noisy, but there is no
systematic method of raw data
grooming that can efficiently reduce this
noise.
At this stage I use gradient smoothing,
limiting the absolute values of the first
order differential of the data (gradient
limiting) to smooth the data
Another look at that BGP
data:…
Gradient Filtered Data
1230000000
1180000000
1130000000
1080000000
1030000000
980000000
Nov-99
Mar-00
Jul-00
Nov-00
Mar-01
Jul-01
Nov-01
Mar-02
Jul-02
Nov-02
Mar-03
Jul-03
Another look at that BGP
data:…


At this stage further smoothing is
necessary in order to reduce the data
set to allow projection models to be
generated
The technique used is a sliding window
average, with a window of 1501 entries
Another look at that BGP
data:…
Smoothed Average
1230000000
1180000000
1130000000
1080000000
1030000000
980000000
Nov-99
Mar-00
Jul-00
Nov-00
Mar-01
Jul-01
Nov-01
Mar-02
Jul-02
Nov-02
Mar-03
Jul-03
Another look at that BGP
data:…
Smoothed Address Advertisement Data
1230000000
1180000000
1130000000
1080000000
1030000000
980000000
Nov-99
Mar-00
Jul-00
Nov-00
Mar-01
Jul-01
Nov-01
Mar-02
Jul-02
Nov-02
Mar-03
Jul-03
Another look at that BGP
data:…


Its now possible to apply a best fit
function to the data.
A linear model appears to be the most
appropriate fit:…
Another look at that BGP
data:…
Linear Squares Best Fit
74
72
70
68
66
64
62
60
58
Nov-99
Feb-00
May-00
Aug-00
Nov-00
Feb-01
May-01
Aug-01
Nov-01
Feb-02
May-02
Aug-02
Nov-02
Feb-03
May-03
Aug-03
Another look at that BGP
data:…



An inspection of the first order differential of the data
reveals why the linear fit is considered to be the most
appropriate for the available data.
While the period through 2000 shows a pattern of an
increasing first order differential that is consistent with
an exponential growth model, subsequent data
reveals an oscillating value for the differential, and a
linear model provides a constant first order
differential
This linear model appears to be the most
conservative fit to the data, although the most recent
data shows a highly visible slowdown in the rate
Another look at that BGP
data:…
daily rate of change in address growth per month
0.016
0.014
0.012
0.010
0.008
0.006
0.004
0.002
0.000
Dec-99
Mar-00
Jun-00
Sep-00
Dec-00
Mar-01
Jun-01
Sep-01
Dec-01
Mar-02
Jun-02
Sep-02
Dec-02
Mar-03
Jun-03
Combining the Data
IPv4 Address Space
140
120
100
80
IANA
RIR
BGP
60
40
20
0
Jan-83
Jan-85
Jan-87
Jan-89
Jan-91
Jan-93
Jan-95
Jan-97
Jan-99
Jan-01
Jan-03
Recent Data
IPv4 Address Space
140
120
100
80
IANA
RIR
BGP
60
40
20
0
Dec-99
Mar-00
Jun-00
Sep-00
Dec-00
Mar-01
Jun-01
Sep-01
Dec-01
Mar-02
Jun-02
Sep-02
Dec-02
Mar-03
Jun-03
Holding Pools:
RIR Pool and Unannounced
Space
Size of Holding Areas (/8)
50
45
40
35
30
UnAnn
RIR
25
20
15
10
5
0
Nov-99
Feb-00
May-00
Aug-00
Nov-00
Feb-01
May-01
Aug-01
Nov-01
Feb-02
May-02
Aug-02
Nov-02
Feb-03
May-03
Modelling the Process



Assume that the RIR efficiency in allocation
slowly declines, so that the amount of RIRheld space increases over time
Assume that the Unannounced space shrinks
at the same rate as shown over the past3
years
Assume an exponential best fit model to the
announced address space projections and
base RIR and IANA pools from the
announced address space projections, using
the above 2 assumptions
Modelling the Process
IPv4 Model
IANA Pool Exhaustion 2022
RIR Pool Exhaustion 2024
200
IANA
RIR
BGP
IANA-P
RIR-P
BGP-P
RIR
LIR
150
100
50
0
Jan-00
Jan-05
Projections
Jan-10
Jan-15
Jan-20
Jan-25
Modelling the Process



Assume that the RIR efficiency in allocation slowly
declines, then the amount of RIR-held space
increases over time
Assume that the Unannounced space shrinks at the
same rate as shown over the past3 years
Assume linear best fit model to the announced
address space projections and base RIR and IANA
pools from the announced address space projections
Modelling the Process
IANA Pool Exhaustion 2030
IPv4 Model
RIR Pool Exhaustion 2037
200
IANA
RIR
BGP
IANA-P
RIR-P
BGP-P
RIR
LIR
150
100
50
0
Jan-00
Jan-05
Jan-10
Projections
Jan-15
Jan-20
Jan-25
Jan-30
Jan-35
Jan-40
Jan-45
Observations



Extrapolation of current allocation practices and
current demand models using an exponential growth
model derived from a 2000 – 2003 data would see
RIR IPv4 space allocations being made for the next 2
decades, with the unallocated draw pool lasting until
2022 - 2024
The use linear growth model sees RIR IPv4 space
allocations being made for the next 3 decades, with
the unallocated draw pool lasting until 2030 – 2037
Re-introducing the held unannounced space into the
routing system over the coming years would extend
this point by a further decade, prolonging the useable
lifetime of the unallocated draw pool until 2038 –
2045
Questions



Will the routing table continue to reflect allocation
rates (i.e. all allocated addresses are BGP
advertised)?
Is the model of the unadvertised pools and RIR
holding pools appropriate?
Externalities:



What are the underlying growth drivers and how are
these best modeled?
What forms of disruptive events would alter this
model?
What would be the extent of the disruption (order of
size of the disruptive address demand)?