Two Million IDNs Tech Day | ICANN 64 | Kobe John Levine | Standcore - - PowerPoint PPT Presentation

▶
two million idns
SMART_READER_LITE
LIVE PREVIEW

Two Million IDNs Tech Day | ICANN 64 | Kobe John Levine | Standcore - - PowerPoint PPT Presentation

Nov 12, 2023 343 likes •540 views

Two Million IDNs Tech Day | ICANN 64 | Kobe John Levine | Standcore LLC | john.levine@standcore.com The contracted zones Number of zones: 1232 Number of names: 193M Number of IDNs: 2M Or about 1% of names The contracted zones

slide-1
SLIDE 1

Two Million IDNs

Tech Day | ICANN 64 | Kobe John Levine | Standcore LLC | john.levine@standcore.com

slide-2
SLIDE 2

The contracted zones

  • Number of zones: 1232
  • Number of names: 193M
  • Number of IDNs: 2M

○ Or about 1% of names

slide-3
SLIDE 3

The contracted zones

slide-4
SLIDE 4

The contracted zones

slide-5
SLIDE 5

What did I do?

  • Get all the zone files, via FTP or CZDS
  • Run them through a script

○ Takes about 90 minutes

  • Put statistics in a database
  • Find out when interesting names were registered

○ Not redacted but still pretty screwed up

slide-6
SLIDE 6

What did I check?

  • Is the name valid under IDNA 2003?
  • Is the name valid under IDNA 2008?
  • Is the name valid under the TLD’s label generation

rules?

slide-7
SLIDE 7

Label Generation Rules: the Theory

  • TLD says what languages it accepts in contracts
  • TLD sends language rules to IANA in RFC3743 format

○ See https://www.iana.org/domains/idn-tables

  • Registered names follow rules for a language/script
slide-8
SLIDE 8

Label Generation Rules: the Practice

  • Some TLDs register languages not in the contract

○ Notably Chinese

  • Table files rarely follow 3743 syntax

○ Files are text but some files in HTML just because

  • Some TLDs haven’t sent in files
slide-9
SLIDE 9

Dealing with LGR files

  • Ad-hoc parser

○ Handles all the files, I think

  • Turn each file into a set of character codes
  • Make merged set of all languages in a TLD
  • Doesn’t capture context dependent rules
slide-10
SLIDE 10

Checking the IDNs

  • Check if valid under IDNA2003
  • Check if valid under IDNA2008
  • See if characters in name are in merged set for TLD
  • If so, see if in merged set for a language in the TLD

○ Not quite right, doesn’t do context rules ○ If TLD changed the rules, doesn’t check date

slide-11
SLIDE 11

What did I find?

  • Most names are OK:

○ Valid under both IDNA2003 and IDNA2008 ○ Matches some language rules

  • Thousands of names are not

○ 509 names invalid IDN2003 ○ 4845 names invalid IDN2008

slide-12
SLIDE 12

Invalid IDN2003

  • 509 names invalid IDN2003
  • All appear to be new valid 2008

○ Many German with ß ○ Arabic with digits

slide-13
SLIDE 13

Invalid IDN2008

  • 4845 names invalid IDN2008
  • Most are length errors
  • About 1000 other errors

○ Pre-2008 legacy junk names ○ New gTLD naughtiness

slide-14
SLIDE 14

Ancient junk

  • Pre-2008 legacy junk names
  • Funky punctuation

○ §sex.com, €-bank.com, 1000°.com, …

  • All appear to be parked or for sale
  • Some look evil

○ xn--google-36d.com g̱oogle.com

slide-15
SLIDE 15

Severe sloppiness

  • Names in non-contracted languages so no LGR

○ 600 Chinese name in .CLUB

  • Not following IDNA rules

○ xn--taylorswift-nu5j.tokyo taylor・swift.tokyo

slide-16
SLIDE 16

Silly games

  • xn--gx5aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

aaaaaaaaaaaaaaaaa.top

  • 頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂

頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂頂 頂頂頂頂頂頂頂頂.top

slide-17
SLIDE 17

Silly games

$ whois xn--gx5aaaaaaa...aaa.top The queried object does not exist: Name is reserved by Registry Primary IDN: xn--gx5aaaaaaa...aaa.top $ host xn--gx5aaaaaaa...aaa.top xn--gx5aaaaaaa...aaa.top has address 23.234.27.100

slide-18
SLIDE 18

Summary

  • Most IDNs are OK
  • Long tail of old junk
  • Some new TLDs need compliance help