APNIC Member Meeting Seoul, Korea, 3 March 2000

APNIC Open Address policy Meeting

Discussion Paper: Minimum IP allocation size

Problem definition

Should APNIC lower the size of the minimum practical allocation from /19 to /20?

1. Motivation

The size of the minimum allocation within the Asia Pacific region being reviewed in response to membership requests, which have mostly arisen in direct response to the ARIN decision to lower the size of their minimum allocation.

In addition, it is also appropriate to periodically examine the size of the minimum allocation in the context of any changes in the technological and administrative environment, as required by APNIC-076, "Policies for address space management in the Asia Pacific region" (http://www.apnic.net/docs/add-manage-policy.html).

Specifically, we must ask whether address space be conserved by lowering the minimum allocation. The wider context of this question is whether it is actually necessary to conserve address space.

2. Background

Historically, APNIC has had different minimum allocation sizes in response to the changing technical environment within which address space is allocated.

Until Jan 1997 the minimum practical allocation was a /22. This was changed by a vote of the membership, which is recorded in the following extract from the member meeting minutes (ftp://ftp.apnic.net/apnic/meetings/Jan97/minutes):

"It was proposed that APNIC modify its allocation policy to more closely match the RIPE-NCC allocation policy. More explicitly, when an organization requests an initial block of address space, APNIC will allocate an initial /19.
This vote was carried : Yes - 65, No - 2"

3. Current status

APNIC
The current minimum practical allocation for APNIC members is a /19 (8,192 addresses). The implementation of this is reflected in the 'slow start' mechanism APNIC and NIRs apply to all new LIRs. With respect to allocations made by APNIC, the first allocation an LIR receives will be the size of the minimum practical allocation. The /19 minimum has not been formally reviewed in more than two years.

ARIN
In October 1998, ARIN members voted to lower the size of the minimum allocation size from /19 to /20, following comments that /19 was too restrictive. (Now, to qualify for an allocation directly from ARIN, an organisation must be multihomed, and have used a /21 from an upstream provider, which they then agree to return once they have received a /19.)

RIPE NCC
The RIPE NCC currently allocate a minimum slow start of /19. In October 1998, the Local IR working group considered lowering the minimum allocation based on the findings presented by the RIPE NCC.

The number of LIRs in the RIPE NCC service region that had not used more than a /20 from their initial /19 allocation within one year was determined. The results were presented as follows:

Of 1,410 LIRs:

304 (22%) had assigned not more than a /20 during the first 12 months of operations;
the total amount of allocated but, after one year, not yet assigned address space was equivalent to 19 /16s;
32 (10%) of the LIRs which had not assigned more than a /20 (32/304) after one year have since become inactive and the address space concerned will not be assigned in the foreseeable future.

These results have been presented to the LIR Working Group (LIR WG) at the 31st RIPE Meeting in Edinburgh. The LIR WG is the body defining local address space policy for the RIPE NCC. As they reported, "the LIR WG didn't feel that this is a reason to change the current allocation policy. The benefits for conservation are not significant". [Source - https://www.ripe.net/ripe/maillists/archives/lir-wg/index.html]

4) Discussion

This paper is an examination of the data correlated from APNIC records and is being presented to the members for their consideration. In each set of statistics gathered, open and current members only were considered. With closed members it is difficult to determine with any accuracy what the usage rates are.

i) Total allocations per member

Prefix*  # Members  % of Membership
----------------------------------
24    0      0.00%
23    2      0.61%
22    10     3.03%
21    6      1.82%
20    8      2.42%
19    188   56.97%
18    43    13.03%
17    26     7.88%
16    19     5.76%
15    15     4.55%
14    5      1.52%
13    2      0.61%
12    0      0.00%
11    2      0.61%
10    1      0.30%
9     2      0.61%
8     1      0.30%
330 100.00%
331
[*The prefix figure is calculated by finding all parties who have received an allocation
(330 in total, excluding non-members and ex-members), counting their total # of /24s
received, and then finding the equivalent single prefix. eg 32x/24=/19, but 48x/24=/18.]

In summary, approximately 65% of the members have a /19 or less, and approximately 35% have used greater than a /19.

The figure of 188 who hold a total of a /19 is weighted with those members who have received allocations within the last year. For this reason the data was examined further to separate out those members. This data is presented below.

ii) Members using more than a /19 than a year compared to those using less than or equivalent to a /19

The set of 312 current and active APNIC members was divided into 3 subsets for analysis, as follows:

A. 84 members having received 2 or more allocations
B. 182 members having received 1 allocation only
C. 46 members having received no allocation so far

Set A represents a small proportion of APNIC members, but is the easiest to analyse as a consumption rate can be established for the time between the first and most recent allocations. This set was therefore able to be further subdivided as follows:

A1. 37 members having consumed more than /19 per year
A2. 47 members having consumed less than or equivalent to /19 per year

Set B includes a greater number of members, but as they have received only one allocation, it is not possible to establish an accurate address consumption rate. However, it is possible to establish a maximum consumption rate for each of these members (based on the number of days since the allocation was made), and to divide the set as follows:

B1. 63 members having a maximum consumption rate of less than or equivalent to /19 per year
B2. 119 members having a maximum consumption rate of more than /19 per year

In the case of Set B2, these members have received an allocation recently, so that is it not possible to determine whether or not they will exceed a consumption rate of /19 per year.

From this analysis is it possible to draw a definite conclusion for 147 of 312 active members (excluding 119 members from set B2). This set of members can then be divided as follows:

1. 37 members having a consumption rate of more than /19 per year (set A2)
2. 110 members having a consumption rate of less than or equivalent to /19 per year (sets A1 + B1)

Without further information therefore, we may draw the conclusion that approximately 75% of APNIC members (i.e. 110/147) are consuming address space at a rate of less than one /19 per year.

iv) Will conservation be improved?

From the above data it appears that by lowering the minimum allocation, conservation would be improved, as only 25% of APNIC members using more than a /19 per year.

v) Is conservation an issue?

Current estimates indicate that approximately 40% of IPv4 address space is allocated (excluding the class B space), and the current rate of consumption globally is roughly estimated at around 3.5 /8's per year.

The widespread deployment of new IP address intensive technologies like ADSL, cable, mobile and wireless networks, where one IP address or more is typically assigned to one user, will affect the rate of address space consumption. With a world population at 6 billion and a limit of 4 billion IPv4 addresses, conservation is clearly still very important.

The rate at which IPv6 is adopted is relevant to this debate; however at this stage it remains unknown.

v) Will lowering the minimum allocation add an increased load on the routing table?

The new policy would not necessarily increase the load on global routing, because the same amount of allocations would still be given out, but with longer prefixes. However, management of the address space could become more complex, due to the finer breakdown of that actual address space.

If as a result of lowering the minimum allocation, more organisations received address space, then the number of prefixes would increase correspondingly.

vi) ISP filtering

Any change to the size of the minimum allocation would need to be well publicised so that filters set by ISPs can be modified accordingly. This would include announcements to major operational mailing lists, such as NANOG, as well as regional lists like apops, apnic-announce, and ISP user communities within specific countries.

5. Recommendations

Conservation could be improved almost immediately if the minimum allocation to all new members was reduced from a /19 to a /20, therefore APNIC recommends lowering the minimum allocation.

Top | SIGs