LIR Survey Results (Supporting data for “Application of ...

1

LIR Survey Results(Supporting data for “Application of

the HD ratio to IPv4” proposal)

Policy SIG 8 Sep 2005

APNIC20, Hanoi, VietnamSave Vocea

2

Why an LIR survey?

• Application of the HD ratio to IPv4 [prop-020-v001]–Feedback that 80% utilisation is difficult to reach

• Replace fixed 80% with variable utilisation (HD ratio)

–Presented at APNIC18• http://www.apnic.net/docs/policy/proposals/

–No clear support or disagreement with proposal• Action on secretariat

–“pol-18-001: Secretariat, with assistance from NIRs,to conduct a survey of ISPs' resource managementpractices to allow a better understanding of issues”

• Motivation–To provide a better service to members

3

Recap…• HD ratio states

–Increasing hierarchy in network leads todecreasing efficiency in addressing

–HD ratio value matches % utilisation whichdecreases as size of address space grow

)log(

)log(

addresses total

addresses host utilised=HD

4

LIR survey questions

5

Details of LIR survey• Design phase

– Consulted network operators (APNIC19, by phone)• Qualitative not quantitative

– Face to face interviews– Conducted with assistance of NIRs and APNIC

training team• Many thanks to both

• Opportunity to ask “extra” questions– NAT, IPv6

• Responses– 67 respondents in total– 15 different economies– Profile reflected that of APNIC membership

6

Survey summary

12%(45%)

8(30)

IPv6 Deployed (and at least planned)

40%27Use of NAT

67Number of responses

15Economies Represented

40%27Members experiencingproblems with 80% policy

7

Methodology for analysis• Use ‘hierarchy’ measures as key to HD

impacts– If trends show relationship with hierarchy then very

likely that HD ratio addresses this– Focus on 80% issues respondents

• Suggests applicability of HD approach

• Considered– Existing use of IPv6 and NAT– Member tier– Address management models

• Service type offering, geographic location (PoP),technology type

8

Member categories surveyed

0

5

10

15

20

25

30

35

Very small Small Medium Large Very large Extra large

Membership category

Num

ber o

f ins

tanc

es

All respondantsRespondants w ith 80% problems

9

Responses by member economy

0

1

2

3

4

5

6

7

8

9

10

11

12

AS AU CK CN FJ ID IN JP KR PH PK TH TW VN VU

Economy

Num

ber o

f res

pons

es

All respondantsRespondants w ith 80% problems

10

Number of PoPs

0

2

4

6

8

10

12

14

16

1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50

Number of PoPs

Freq

uenc

y

All respondantsRespondants w ith 80% problems

11

Service categories

0

5

10

15

20

25

30

35

Dial up Broadband IP phones Webhosting

Co location IDC Gaming Wireless Lease line DSL

Service type

Num

ber o

f res

pons

es

All respondantsRespondants w ith 80% problems

12

Number of service categoriesoffered

0

2

4

6

8

10

12

14

16

18

20

1 2 3 4 5

Number of services offered

Freq

uenc

y

All respondantsRespondants w ith 80% problems

13

Address distribution models

0

5

10

15

20

25

30

1 – Geographic location

2 - Customer type 3 - Product 4 - Others

IPv4 address distribution model

Freq

uenc

y

All respondantsRespondants w ith 80% problems

14

No. of address distribution modelsused

0

5

10

15

20

25

30

1 2 3

Number of IPv4 address deployment models used

Freq

uenc

y

All respondantsRespondants w ith 80% problems

15

Types of NAT use

0

5

10

15

20

25

30

35

40

Don't use NAT Infrastructure Customer network

All respondantsRespondants w ith 80% problems

16

Reasons for NAT use

0

5

10

15

20

25

30

35

40

Conservation Security Lack of IP’s Customersevice

Policy issues Don’t use

Reasons for NAT use

Cou

nt

All respondantsRespondants w ith 80% problems

17

Service types vs hierarchy

Weak trend: more services types implies more hierarchy

0

1

2

3

4

5

6

0 1 2 3 4 5 6

Number of service categories offered

Hie

rarc

hy

18

PoPs vs hierarchy

Weak trend: more PoPs require more hierarchy

0

1

2

3

4

5

6

0 10 20 30 40 50 60 70 80 90 100

Number of PoPs

Hie

rarc

hy

19

Member size vs hierarchy

No strong trend. All member-sizes have range of hierarchies

0

1

2

3

4

5

6

0 1 2 3 4 5 6

Member size

Hie

rarc

hy

20

Address deployment vs hierarchy

No Trend. Range of address deployment models used

0

1

2

3

4

5

6

0 1 2 3 4

Number of IPv4 address deployment models used

Hie

rarc

hy

21

Conclusion from survey• Total of 40% reported problems reaching

80% utilisation• No correlation between problems and

network size or complexity–Measured as

• No. of PoPs• No. of services deployed• No. of levels of hierarchy

22

Next steps?• Do we need to widen the sample size?• Should this proposal cease?• Continue discussions on the list?• Wait and see - situations in other RIRs

23

Questions?

Thank you!