An Analysis of IPv6 Prefix Announcements in Our Region

Introduction

The purpose of this study is to present a summary of the visibility status of IPv6 prefixes in our region.

Rationale

We have identified a large number of organizations in Latin America and the Caribbean that have not yet announced their IPv6 prefixes despite having received one. We hope this analysis will create awareness among these organizations so that they will take the first step towards their utilization.

Data Sources

The data was taken from the following sources:

  • LACNIC's delegated extended [1] for IPv6 prefixes
  • RIPE's Routing Status API [2] for the status of IPv6 prefixes in the DFZ

Procedure

The process is quite linear: basically, the status of each of LACNIC's IPv6 prefixes on the global routing table was determined.

Examples:

** Case 1: The assigned prefix is identical to the one found on the global routing table: **

https://stat.ripe.net/data/routing-status/data.json?resource=2803:6680::/32

The result is JSON code that shows the origins for the queried prefix resource, which means that the prefix announced on the BGP table is identical to the one assigned by LACNIC.

    _data_

    _query_time "2020-01-17T16:00:00"_

    _resource  "2803:6680::/32"  ← Prefijo anunciado = prefijo asignado por LACNIC_

    _origins_

    _0  ← ID para cada origins_

    _origin    267789   ← AS que hace el anuncio_

** Case 2: The assigned prefix is announced on more specific subnets (an announcement that is not identical to the one that was assigned).

https://stat.ripe.net/data/routing-status/data.json?resource=2001:13c7:6010::/44

In this case, resource appears, but without any origins:

    _data_

    _query_time "2020-01-17T16:00:00"_

    _resource  "2001:13c7:6010::/44"_

    _origins   []   ← Vacio_

In addition, in the JSON code below there is a section titled more_specifics that clearly shows which subnets of the original prefix are being announced.

    _more_specifics_

    _0  ← ID 0_

    _origin    52375  ← AS origin de una subred del prefijo orignal_

    _prefix    "2001:13c7:6011::/48" ← Una subred dentro del prefijo original_

    _1_

    _origin    52404  ← AS que hace el anuncio de una subred del prefijo orignal_

    _prefix    "2001:13c7:6013::/48"  ← recordar el prefijo original era: /44_

    _2_

    _origin    52500  ← AS que hace el anuncio de una subred del prefijo orignal_

    _prefix    "2001:13c7:6014::/48"_

** Case 3: The assigned prefix is not being announced **

https://stat.ripe.net/data/routing-status/data.json?resource=2801:80:1d40::/48

Neither the origins of the original prefix nor the more_specifics section contain any information:

    _data_

    _query_time "2020-01-17T16:00:00"_

    _resource  "2801:80:1d40::/48"_

    _origins   []  ← Vacio_

    _less_specifics    []_

    _more_specifics    [] ← Vacio_

Other Cases

Other cases may exist, such as a prefix that is ??announced in its entirety when there are other, more specific announcements. However, these cases are not taken into account because they do not contribute to the purpose of the study.

Data Processing

The information was processed and debugged using Python 3.

Results Verification

At least two dozen random results were selected and compared with different looking glasses on the Internet, mainly Hurricane Electric's looking glass [3].

Term of the Study

From 16 to 18 January 2020.

Population and Sample

9,781 IPv6 prefixes assigned by LACNIC as at 16 January 2020
2 Prefixes were removed as the country was stated as US
9,749 were successfully processed

Results (only countries for which at least 5 prefixes were obtained and processed are shown):

References:

C=Anuncio completo (prefijo asignado = prefijo visto en la DFZ (Default Free Zone o Tabla de enrutamiento Global))

AP=Anuncio parcial (prefijo asignado no se observa en la DFZ sin embargo <span style="text-decoration:underline;">al menos</span> una subred más chica sí)

NA=No anunciada (prefijo asignado ni prefijos más chicos de la misma se observan en la DFZ)

Total=Total de prefijos IPv6 del país

%AC = Porcentaje de AC respecto al Total

%AP = Porcentaje de AP respecto al Total

%NA = Porcentaje de NA respecto al Total

PAIS  AC AP NA TOTAL  %AC %AP %NA
AR

162

28

742

932

17.38

3

79.61

BO

7

2

30

39

17.95

5.13

76.92

BR

3309

300

3395

7004

47.24

4.28

48.47

BZ

7

2

18

27

25.93

7.41

66.67

CL

39

11

203

253

15.42

4.35

80.24

CO

110

29

162

301

36.54

9.63

53.82

CR

40

10

51

101

39.6

9.9

50.5

CU

2

0

3

5

40

0

60

CW

7

0

8

15

46.67

0

53.33

DO

27

3

24

54

50

5.56

44.44

EC

36

19

98

153

23.53

12.42

64.05

GF

1

0

4

5

20

0

80

GT

18

4

21

43

41.86

9.3

48.84

HN

23

1

78

102

22.55

0.98

76.47

HT

3

0

6

9

33.33

0

66.67

MX

47

31

190

268

17.54

11.57

70.9

NI

9

1

12

22

40.91

4.55

54.55

PA

18

6

45

69

26.09

8.7

65.22

PE

17

7

65

89

19.1

7.87

73.03

PY

33

4

33

70

47.14

5.71

47.14

SV

10

2

26

38

26.32

5.26

68.42

SX

3

0

3

6

50

0

50

TT

3

1

7

11

27.27

9.09

63.64

UY

10

2

16

28

35.71

7.14

57.14

VE

32

3

59

94

34.04

3.19

62.77

Table 1

Results Summary

  • The countries with the highest rate of complete prefix announcements (AC by their Spanish acronym) were Saint Maarten and the Dominican Republic (50%), followed by Brazil (47.24%).
  • The country with the highest rate of partial announcements was Ecuador (12.42%). Given the nature of this indicator, it should be noted that were not expecting these numbers to be high. However, understanding the reason behind this may be of interest.
  • We can see that the country with the highest rate of unannounced IPv6 prefixes was CL (80.24%), followed by French Guiana (80%) and Argentina (79.61%).
  • Finally, by applying a different perspective, we can determine which country has the highest rate of prefix announcements by taking a look at the %NA column. In this case, the country with the highest rate of announcements is Dominican Republic, for which %NA is 44.44%, meaning that 55.56% of their prefixes are visible in the DFZ (congratulations!).

Conclusion

The number of IPv6 prefix announcements for the entire region is less than 50% of the number of assignments. There is still a long way to go. Knowing the countries where the difference between the number of assigned vs announced prefixes is the highest may help us understand which locations will require more of our time. We must continue to work to reduce the difference between assignments and announcements.