Naming IPv6 address parts – Proposed IETF RFC and a VOTE

Some time ago I posted a word “chazwazza” to the Network Dictionary :Network Dictionary  Chazwazza. Although it was mostly intended to be a humourous comment, what makes it funny is that we don’t actually have a good naming convention for the ‘chazwazza’ of an IPv6 address. Apparently I’m not the only person to have noticed.

L. Donnerhacke and Richard Hartmann have created an RFC draft-denog-v6ops-addresspartnaming-02 – Naming IPv6 address parts that is preceding to Proposed Standard. However, we need to develop a mechanism to provide some evidence of a preferred name for the IETF to assist them in making a recommendation and finalise the voting process. Therefore, to start the process I have setup the UNOFFICIAL IPv6 NAMING POLL. Please join in the voting so that we can forward the results to the RFC Proposers as part of the evidence gathering process. Here is your chance to part of the Internet history.

My personal preference (all jokes aside) is for the term CHUNK. It’s easy to say, a single syllable and fairly descriptive. This i can easily imagine saying:

So in the forth chunk, it’s EECA, and the fifth chunk id FFEI

Background

An IPv6 address is 128 bits long, and written as 32 hexadecimal characters. Like so:

Ipv6 naming 1

By convention, we represent the IPv6 addresses with colons between each 4 hexadecimal characters to make it easy to represent, read and identify because human short term memory can readily cope with ‘chunks’ of three or four digits as demonstrated by telephone numbers. A non-delineated string of 32 characters is very hard for the human mind to comprehend. This is covered in RFC 2373.

And there are other purposes (note that the I in Caffeine is a 1) : Ipv6 naming 2

It’s the current convention to use the right most 64 bits as the “Interface ID”.

IPv4 Terminology

For IPv4, we used the term OCTET to describe the three digit chunk of an address. Ipv6 naming 3 This diagrams show the FIRST OCTET as 198.

What do we name each section of an IPv6 address ?

Ipv6 naming 4

Since we are likely to spend the rest of our working lives working with IPv6, we need a convenient handle.

Naming the IPv6 Address Parts

The standard proposes the following names:

Term Description
4.1. Chazwazza “Chazwazza” was proposed as a Simpsons reference, see [greg]. While this is certainly a unique term in the networking world, it is not particularly meaningful nor easy to pronounce.
4.2. Chunk A chunk is commonly understood to be a specific amount of data. The term is not unique to IPv6, however easy to remember and pronounce.
4.3. Column The colons in an IPv6 address’ text representation make it similar to a table. Besides that, the meaning of the word “column” has very little to do with the actual technical meaning of a 16bit piece of an IPv6 address, though.
4.4. Colonade, Colonnade Based on the colon as seperator the word sounds English (using a single ‘n’ to make it an artificial word) and is easy to spell and pronounce.Alternatively, “colonnade” could be used, overloading the existing, yet unrelated word with a new meaning.
4.5. Doctet Derived from “double octet”, thus accurately describes the technical matter, as an octet is a standard term for a sequence of 8 bits.
4.6. Field A “field” describes a form of a data structure in many programming languages. The term stresses the fact that a field is one of multiple fractions of a bigger subject, just like countryside is divided into fields, or like IPv6 addresses into 16bit long pieces. A drawback of that similarity is the lack of uniqueness to IPv6, though.
4.7. Hexadectet “Hexadectet” is directly derived from IPv4’s “octet”, thus technically correct and probably convenient to get used to. On the other hand, it is much harder to pronounce.
4.8. Hit Short for “hex-bit”, short and convenient to pronounce, however usually associated with a completely different meaning.
4.9. Orone Initially started as a typo in [greg], “orone” is a short, unique word without a specific meaning yet.
4.10. Part The word “part” has been used throughout this document to describe the subject until there is a better term for this. It is very unspecific and can be used in countless ways, not only to describe 16bit long parts of an IPv6 address.
4.11. Provider number, customer number, network number These terms provide semantic descriptions of the different parts of an IPv6 address. However, it is not within the scope of this document to find terms describing semantic, but rather syntactic elements.Furthermore, naming the 16bit pieces of IPv6 addresses in a semantic way would introduce new problems, like limited applicability, e.g. it would not work for multicast addresses.
4.12. Quad nibble, qibble, quibble A nibble is a 4bit entity, hence 16 bits are a quad nibble. This is a rather bulky word, however, so “quibble” is a convenient abbreviation. Also, it is a unique term, thus eliminating any chances of misinterpretation.
4.13. Segment “Segment” is another obvious choice, however it is also quite unspecific and used in different contexts, e.g. “network segments”.
4.14. Tuple A tuple is a sequence of typically heterogenous elements considered as a new entity by itself. It is also a short, descriptive word that is not yet associated with anything networking related. Usually a tuple exceeds grouping by creating a new semantic level.
4.15. Word A “word” usually refers to a fixed group of bits that are processed at a time, and especially on legacy x86 systems is a synonym for 16 bits. It has a different and much more unspecific meaning to less technically skilled people, which might be problematic.

POLL

[poll id=”12″]

  • http://perlmonkey.blogspot.com/ Russell Heilling

    My serious answer “field”. For comedy value: “quibble”

    quibble (?kw?b ? l)

    ó vb
    1. to make trivial objections; prevaricate
    2. archaic to play on words; pun

    Seems apt to me…

    • http://www.cisco.com noah

      I like “quad” (short for quad nibbles)

  • Tristan Rhodes

    I really like the term “doctet”. Most people didn’t know why an “octet” was called an “octet”, but they used it all the time. Doctet will build on the previous familiarity of “octet”, and most network engineers will just adopt it without thinking. In fact, I am going to start using “doctet” right now.

    Thanks for your valuable contributions!

    Tristan

  • http://www.iks-jena.de/mitarb/lutz/ Lutz Donnerhacke

    We have an ‘offical’ poll on this issue: http://doodle.com/5q9gfvk4qe6zmzc6
    Current leading term is “hextet” (for good reasons)
    Some weeks ago it was not that clear: http://wwwneu.iks-jena.de/eng/Blog/Talking-about-IPv6

    • http://etherealmind.com Greg Ferro

      Lutz

      Thanks for the update. However, IMHO hextet is a ‘crap’ sounding word to me. It’s kind of pretentious and difficult to say. And that poll looks heavily self selecting but lets see what the results from this (pseudo-independent) pool look like and compare.

      I will add HEXTET to the poll choices.

      greg

  • Colonnade Boston

    We’re partial to “Colonnade” – it just sounds luxurious and hip. :-)

  • http://www.netresec.com/ Erik

    Great blog post Greg!

    I’m spreading the word about your poll here:
    http://www.netresec.com/?page=Blog&month=2011-02&post=Name-the-Chazwazza-in-IPv6

    But I actually agree with Lutz that “Hextet” would be an even better name since it’s obvious that it is related to the IPv4 “octet”.

  • http://www.wireshark.org/ Gerald Combs

    I humbly propose “chex”, short for “chunk of hexadecimal”.

  • Pingback: Sniff free or die » How To Read IPv6 Addresses()

  • http://www.gtech.com Chris

    Hextet is technically incorrect as it implies 6 bits, not 16, and has already been dismissed in the draft (see Appendix A).

    I think quad would make sense if octets had been referred to as doubles, but we typically don’t count nibbles, although the draft lists quibble to avoid ambiguity.

    Overall, so far I think chunk is better as nibble < byte (i.e., "bite") < chunk, but here are some other possible ideas just to throw some out there …
    1) a short
    2) a slice
    3) a gob
    4) a nugget
    5) a sulfur :)

  • Charles

    “shim”

  • Ivan Brunello

    My two cents, as a non-native English speaker (Italian)

    “Chunk” fits well, since it’s easy to use as-is, even in foreign languages (at least, Italian).
    And seems to have a meaning by itself.

    A short word for a longer composite word (quibble, hit,doctet), would either lead to really ugly translations (read “DOTTETTO” :-( ), or be kept as original and remain almost meaningless.

    A longer word (chawWHAT?) should be unneeded tongue exercise.

    My two cents

    Ivan

    • http://etherealmind.com Greg Ferro

      Ivan

      I agree. Chazwazza was meant to be a joke. As I said, I think chunk is best, but other people are suggesting the use of hextet (since we use octet for IPv4).

      greg

  • Adrianne

    How about “chomp”?

    • http://etherealmind.com Greg Ferro

      Sadly, they ended up approving “hextet” here.

      Which is a bit boring.

  • Loren

    Octet = 8 bits (i.e. 8 binary digits)
    Hextet = 16 bits

  • Loren

    On second thought,  I like Quad-nibble idea expressed earlier.
    While working with hex code, it is easier to look at 4 bits at a time (one hex character).
    And with a group of hex characters, the boundaries break nicely into 4 bits.
    So it seems natural and accurate to represent a colon separated segment of the IPv6 address as four hex-coded nibbles.
    or, simply as one “Quibble”
    Yes, this works very well.

  • mccs

    i would say its a DEX from 
    Hexadectet but this is short and not related to something else

  • http://www.facebook.com/lynne.patterson55 Lynne Patterson

    I teach IPv6 and I use the very technical word, “chunk.” I do, however, love chazwazza. It is fun to say and a great scrabble word as well.

    • http://etherealmind.com Greg Ferro

      Yeah, I know. Fun detectors.

  • Nic Bhasin

    Long time ago in Wendell Odom’s CCNA book (and only there), they were called Quartets. I have always liked that.