Slashdot Mirror


To Keep Track of World's Data, You'll Need More Than a Yottabyte (wsj.com)

An anonymous reader shares a report: In 10 or 15 years, Dr. Brown, who is head of metrology at the National Physical Laboratory in the U.K., anticipates the amount of computerized data worldwide will exceed 1 yottabyte in size, and without expanding the list of prefixes, there will be no way to talk about the next great chunk of numbers. Even worse, dilettantes could fill the void by popularizing glib prefixes such as bronto or hella -- terms that have already won fans. Without professional intervention, Dr. Brown fears, the next numerical prefix could become the Boaty McBoatface of weights and measures.

[...] For the record, there is an argument to be made for adopting a prefix like bronto: giga and tera are based on the Greek words for "giant" and "monstrous." Why not make bronto, named for the brontosaurus, official, perhaps along with tyranno, stego, colosso or even yeti? Dr. Brown is sympathetic to the argument but unconvinced. Instead, he proposes four prefixes that adhere to recent naming conventions [Editor's note: the link may be paywalled; an alternative source was not available.]: ronna and quecca for octillion (27 zeros) and nonillion (30 zeros), along with ronto and quecto for their fractional counterparts, octillionth and nonillionth. Like the latest sanctioned prefixes, Dr. Brown's proposals are loosely related to Latin and Greek words for numbers (in this case, nine and 10). And like most of the prefixes, his suggestions end in "a" or "o." But the process of expanding, or even amending, the official measurements is lengthy.

8 of 81 comments (clear)

  1. Somewhat arbitrary what we call data by ganv · · Score: 2

    It has always seemed a bit arbitrary to label something as "the world's data". You could always add the history of every cache on every processor on the planet to your definition of "data" and have a much larger number.

    1. Re:Somewhat arbitrary what we call data by ShanghaiBill · · Score: 4, Funny

      It has always seemed a bit arbitrary to label something as "the world's data".

      A yottabyte is 1e24. That is more than 100 terabytes per human.

      You could always add the history of every cache on every processor on the planet to your definition of "data" and have a much larger number.

      640 yottabytes ought to be enough for anyone.

    2. Re:Somewhat arbitrary what we call data by apoc.famine · · Score: 2

      I know, right? I've been backing up /dev/random for years now, and I'm not sure when I'll be done. I think part of the problem is running the checksum, but I'm not sure.

      --
      Velociraptor = Distiraptor / Timeraptor
  2. Fake need? by Tablizer · · Score: 2

    Why make a new prefix for each power of ten unless (and until) it really is used often? Just make a generic term, such as "24th order of magnitude". In fact, I believe that's already used. We can even have a shorthand: "24 oom bytes". To remember it, think of a cow mooing in reverse.

    1. Re:Fake need? by viperidaenz · · Score: 4, Insightful

      They're not.
      they're making new prefixes for every third power of 10.

      If a consensus isn't reached relatively soon, the whole "billion" thing will happen again. it's been defined as both 10 to the 9th power and 12th power.

  3. Byte my shiny metal exponents by az-saguaro · · Score: 3, Interesting

    The names must apply to all forms of measures and metrics.
    But, if the Bureau of Geeks and Nerds has its say, the names will be:

    whata-byte
    abigga-byte
    onthisa-byte
    myassa-byte
    heybitchdont-byte

    On the serious side, the current system requires us to remember three names or prefixes for each triad (each 10^3).
    For example:
    one-million or one-millionth, versus one mega-meter versus one micro-meter. Million-mega-micro-.
    one-thousand or one-thousandth, versus one kilo-liter or one milli-liter. Thousand-kilo-milli-.
    For Europeans and others speaking Latin or Romance languages, the cardinal number names may be closer to the multiplier-divider prefixes, but it is still a cumbersome system.

    For the higher order new numbers, why not make them with a uniform naming convention.
    For instance, the common root name, then tillo- and tetto-.
    Examples:
    10^27 = one octillion trees, one octillo-meter, one octetto-meter.
    10^30 = one nonillion beans, one nonillo-newton, one nonetto-joule.
    10^39 = one dodecillion electrons, one dodecillo-farad, one dodecetto-ohm.

    Instead of having unique initials as abbreviations, such MB, mm, cm, km, Gb, etc., try this, using "D" for "decade":
    My new computer has 4 of 10^27 byte chips = 4-D27B of memory.
    The distance to so and so galaxy is one nonillion meters away, or D30m away.
    Or, something like that.

    It just seems too cumbersome to remember too many contrived names and disparate prefixes for ever bigger numbers that no one can really comprehend or has the time to recall in the middle of a sentence that is meant to be fluent.

    1. Re:Byte my shiny metal exponents by mlheur · · Score: 2

      2^10 = 1KiB
      2^20 = 1MiB
      2^30 = 1GiB
      2^40 = 1TiB
      2^50 = 1PiB
      2^60 = 1EiB
      2^70 = 1ZiB
      2^80 = 1YiB

      now we add...

      2^90 = 1NiB (ninobyte)
      2^100 = 1DiB (decabyte)
      2^110 = 1LiB (levenbyte)
      2^120 = 1WiB (tWelvebyte)
      2^130 = 1BiB (because B looks like 13 in the right font/print)

      or just stop using prefixes and go full maths on it. e.g. "there are 3.250 x 2^98 bytes of storage"

  4. Re:Horseshit by Waffle+Iron · · Score: 3, Insightful

    Well, other than using powers of 1,024 (or powers of 1,000 for the pedantic types who are unfamiliar with base 2.)

    I'm familiar with base 2. So I know that hard drives typically allocate blocks of size 2^9 or 2^12, and I know that there is nothing else in a hard drive related to powers of two.

    Which means that insisting on using powers of 1024 notation is like demanding that we count everything related to the NBA in base 5, since basketball teams have 5 members.

    (Actually, power-of-1024 notation is even worse than that, since it uses a *mixture* of various mutually incompatible 1024 powers combined with decimal fractions, all of which makes Roman Numerals look practical by comparison.)