ipv6%address%populaons%in%the% wild% · outline% • the%experiment • results%and%analysis% –...

12
IPv6 Address Popula0ons in the Wild Manish Karir DHS/S&T/CSD Geoff Huston George Michaelson APNIC Michael Bailey University of Michigan

Upload: others

Post on 01-Aug-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: IPv6%Address%Populaons%in%the% Wild% · Outline% • The%experiment • Results%and%analysis% – Populaon%mix% – Teredo % – 6to4 – MAC%and%EUI64% • Conclusions

IPv6  Address  Popula0ons  in  the  Wild  

Manish Karir DHS/S&T/CSD

Geoff Huston George Michaelson

APNIC

Michael Bailey University of Michigan

Page 2: IPv6%Address%Populaons%in%the% Wild% · Outline% • The%experiment • Results%and%analysis% – Populaon%mix% – Teredo % – 6to4 – MAC%and%EUI64% • Conclusions

Outline  

•  The  experiment  •  Results  and  analysis  – Popula0on  mix  

– Teredo  – 6to4  – MAC  and  EUI-­‐64  

•  Conclusions  

Page 3: IPv6%Address%Populaons%in%the% Wild% · Outline% • The%experiment • Results%and%analysis% – Populaon%mix% – Teredo % – 6to4 – MAC%and%EUI64% • Conclusions

Unique  IPv6  Addresses  per  day  

0

50000

100000

150000

200000

250000

300000

350000

400000

450000

Jan 1 2012

Feb 1 2012

March 1 2012

April 1 2012

May 1 2012

June 1 2012

June 9 2012

July 1 2012

Num

ber o

f Uni

que

IPv6

Add

ress

es

Month

Number of Unique IPv6 Addresses per Day

total impressions

observed unique IPs

Page 4: IPv6%Address%Populaons%in%the% Wild% · Outline% • The%experiment • Results%and%analysis% – Populaon%mix% – Teredo % – 6to4 – MAC%and%EUI64% • Conclusions

IPv6  Popula0on  Mix  

Page 5: IPv6%Address%Populaons%in%the% Wild% · Outline% • The%experiment • Results%and%analysis% – Populaon%mix% – Teredo % – 6to4 – MAC%and%EUI64% • Conclusions

Regional  Distribu0on  of  Na0ve  IPv6  Addresses  

Page 6: IPv6%Address%Populaons%in%the% Wild% · Outline% • The%experiment • Results%and%analysis% – Populaon%mix% – Teredo % – 6to4 – MAC%and%EUI64% • Conclusions

Decoding  Teredo  

•  Encoded  Informa0on:  –  Prefix  2001:0000::/32  –  Teredo  server  –  NAT  flag  –  UDP  Port  –  Client  IPv4  

Source:  Wikipedia  

Page 7: IPv6%Address%Populaons%in%the% Wild% · Outline% • The%experiment • Results%and%analysis% – Populaon%mix% – Teredo % – 6to4 – MAC%and%EUI64% • Conclusions

Teredo  Servers  

•  11M  unique  IPv6  addresses  

•  Overall  258  Teredo  servers  in  171  unique  ASNs  in  39  countries  

•  150  in  RIPE  region,  86  in  ARIN,  19  in  APNIC,3  in  AFRINIC,  none  in  LACNIC  

•   Top  5  servers  account  for  almost  all  of  the  addresses  observed  

Page 8: IPv6%Address%Populaons%in%the% Wild% · Outline% • The%experiment • Results%and%analysis% – Populaon%mix% – Teredo % – 6to4 – MAC%and%EUI64% • Conclusions

Teredo  and  Cone  NAT  

•  15K  unique  IPv6  addresses  had  “cone  NAT”  flag  set  •  Almost  half  in  APNIC  region  –  7.5K,  5K  in  RIPE  

•  Indonesia  and  New  Zealand  Telecom  largest  contributors  

Page 9: IPv6%Address%Populaons%in%the% Wild% · Outline% • The%experiment • Results%and%analysis% – Populaon%mix% – Teredo % – 6to4 – MAC%and%EUI64% • Conclusions

6to4  Usage  by  Country  

0

100000

200000

300000

400000

500000

600000

SA

GE

RS

GR LV

NL

TR IL PT

ES

CZ

SG LT FI

FR

PK

KZ

CA

PH

BG

VE

ZA

DE

HK

SE

MY

MA

BY

BD

CO JP CL

GB

HU PL

AR IN

MD

UA

US IT

RO

RU

TW CN

AU

BR

KR

NZ ID

Num

ber

of U

niq

ue 6

to4 A

ddre

sses

Country

6to4 Usage by Country

•  Prefix:  2002::/16  •  Next  32  bits  are  IPv4  address  

•  2.78M  unique  6to4  addresses  •  RIPE-­‐  92K,  1.4M  –  APNIC,  

112K  –  ARIN,  292K  –  LACNIC,  65K  –  AFRINIC  

•  205  unique  country  codes  •  Indonesia  largest  –  500K  –  

18%  •  New  Zealand,  Korea,  Brazil,  

Australia,  China  ~100K  -­‐  4%  each  

Page 10: IPv6%Address%Populaons%in%the% Wild% · Outline% • The%experiment • Results%and%analysis% – Populaon%mix% – Teredo % – 6to4 – MAC%and%EUI64% • Conclusions

Decoding  MAC  addresses  

Source:  Wikipedia  

•  The  boeom  64bits  of  an  IPv6  address  are  the  host  iden0fier  which  can  be  generated  based  on  the  48-­‐bit  MAC  address    

•  Iden0fied  by  the  tag  “FF:FE”  in  12,13  bytes  

•  Possible  for  both  na0ve  IPv6  as  well  as  6to4  addresses  

•  Host  iden0fiers  can  also  be  generated  via  DHCPv6  or  randomly  

Page 11: IPv6%Address%Populaons%in%the% Wild% · Outline% • The%experiment • Results%and%analysis% – Populaon%mix% – Teredo % – 6to4 – MAC%and%EUI64% • Conclusions

Observed  MAC  Addresses  •  Of  174K  na0ve  IPv6  addresses  only  10%  had  FF:FE  tag  –  13K  

•  Of  2.78M  6to4  addresses  only  9K  had  FF:FE  tag  

•  Apple  vendor  ID  accounts  for  largest  number  of  hosts  iden0fied  close  to  7K  for  each  na0ve  and  6to4  addresses  

Page 12: IPv6%Address%Populaons%in%the% Wild% · Outline% • The%experiment • Results%and%analysis% – Populaon%mix% – Teredo % – 6to4 – MAC%and%EUI64% • Conclusions

Conclusions  

•  14M  IPv6  hosts  •  Mostly  transi0on  technologies  –  Teredo  ~80%  •  Low  usage  of  na0ve  IPv6  ~1.3%  •  Higher  Teredo  usage  in  RIPE  region  •  Significantly  higher  usage  of  6to4  in  APNIC  region  •  Not  seeing  very  many  host  iden0fiers  with  MAC  addresses  

•  Many  more  hosts  IPv6  literal  capable  ~30%  (of  all  IPv6  addresses  observed)