Measuring data storage, SI prefixes,
or "a yotta stuff to think about...."

byte
8 bits ("bit" is short for "binary digit". A single 0 or 1 in a two-state system.)
1 byte can store 1 keystroke.

Half a byte (4 bits) is a "nybble". I'm not joking.

Silly analogy, as a visual aid: one strand of hair has a thickness of about .05mm. Read on...

kilobyte
103 or 1,000 bytes (1 thousand). The actual number is 210 or 1,024 bytes unless you're in the marketing department.

The International Electrotechnical Commission (IEC) has adopted the term "kibibyte" to combat the inexactness caused by the competing definitions of kilobyte (103 or 210). Likewise, they propose that we start using "mebibyte", "gibibyte", and similar variations down the line. (The "bi" refers to the binary basis). Click here for details at nist.gov.

These terms sound silly, however, so we'll present the more common terms here until the rest of the universe jumps on the bandwagon, with the understanding that it could be interpreted as a power of 10 or a power of 2. When you get to billions or more, 1 billion (109) or 1.07 billion (230) are both a heck of a lot of bytes, so we won't be too picky about the difference here on this page. Of course, I'm not trying to sell you a hard disk, or explain where the "missing bytes" went...

A kilobyte is the equivalent of about 2/3rds of a page of text.

In the late 1970's, a 5 1/4" floppy disk held about 150 kilobytes.

Silly analogy, part 2: a "kilohair", or 1000 hairs side by side (very carefully. don't sneeze.), would measure 50mm, or 5cm, or about 2 inches.

megabyte
106 or 1,000,000 bytes (1 million), or 220 (1,048,576) using binary.

1,000 kilobytes.

A megabyte can hold the text of 1 to 2 books in uncompressed format. Check Project Gutenberg to see. Mark Twain's Huckleberry Finn fits in a 563Kb text file. His Following the Equator takes 1.05Mb.

If you were tired of swapping floppies on your desktop computer, in the early 1980's you could buy a 5 megabyte hard disk for just over $2000. It was about the size of a shoebox.

A 3 1/2" floppy disk holds 1.4 megabytes of data. These were commonly available by the mid 1980's.

Silly analogy, part 3: a "megahair", or a million hairs side by side, would stretch 50 meters, or a little more than halfway down a football field. Since the average human has about 100,000 hairs on his or her head, we'd have to find about 10 people who want to try the Michael Jordan look to test this.

gigabyte
109 or 1,000,000,000 bytes (1 billion), or 230( 1,073,741,824) in binary.

1,000 megabytes, or a million kilobytes

One gigabyte can hold the text of over 1,000 books. That's a pretty decent-sized library bookshelf.

A compact disc holds about 3/4ths of a gigabyte. These were also available in the mid-1980's. By the mid-1990's recordable CDs were available for data storage, and by the late 1990's gigabyte hard disks were becoming common.

A gigabyte holds about 100 minutes of CD quality stereo music.

Single-sided DVDs hold from 4.7 gigabytes if recorded in one layer, or about 8.5 gigabytes if double-layered. Computer recordable DVDs are single-layered. Movie DVDs are usually double-layered.

One gigabyte holds about 25 minutes of DVD-quality video. A typical movie requires about 6 gigabytes of storage.

Silly analogy, part 4: a "gigahair", or one billion hairs side by side, be about 50,000 meters, or 50km. That's 30 miles. You'd have to find a town of about 10,000 people who were all willing to get haircuts at the same time if you wanted to try this.

terabyte
1012 or 1,000,000,000,000 bytes (1 trillion), which is slightly less than 240

1,000 gigabytes, or a million megabytes

Over 1,000,000 books. We're talking about a pretty large library here.

All the text in all the printed matter in the Library of Congress would fit in about 20 terabytes.

A terabyte would hold the contents of about 2,000 audio CDs in original uncompressed format.

A terabyte can hold over 160 DVD movies.

If Moore's Law* continues to hold when applied to data storage, we'll have one terabyte storage devices commonly available by 2010. They'll as portable as CDs and DVDs are now.

Silly analogy, part 5: a "terahair", or a trillion hairs carefully laid side by side, would be 50,000 km or 30,000 miles in length. That's 1 1/4 times around Planet Earth at the Equator (circumference 40,070 km). But of course, to try it, you'd have to persuade just about everyone in New York City and Chicago to get haircuts for you. That's about 10 million people times 100,000 hairs each.

petabyte
1015 or 1 quadrillion bytes. In binary, it's about 250.

1,000 terabytes. A million gigabytes. A billion megabytes.

Over one billion books. Are there even that many unique books in the history of printing?

Roy Williams Clickery** estimates 2 petabytes as the contents of all U.S. academic research libraries, and 200 petabytes as "all printed material". But of course those include duplicate copies!

A petabyte could hold about 2 million audio CDs, uncompressed. Compress them to 128kbps MP3 files, and you'll fit about 20 million CDs. This is in the range of all recordings ever made.

160,000 DVD movies would fit in a petabyte.

You could have a nice DVD-quality collection of all films ever released in about 2 petabytes. The Internet Movie Database lists about 250,000 theatrical releases, and about 325,000 including made-for-TV movies and other films.

So in about 4 or 5 petabytes, we fit ALL OF THE ABOVE. All books. All audio recordings. All movies. We can add more petabytes and start to add in all photographs ever taken. Then start adding everything ever broadcast on television...

Who knows when we'll run into physical limitations? But if Moore's Law continues to hold when applied to data storage, we'll have one petabyte storage devices that you can walk around with in your hand by around 2025.

"petahair": side by side, we're now at 50,000,000 km. Or 5x1010 meters. That's about the distance to Mars when it's at its closest position to Earth. It would take the hair of about 10 billion people, but we only have about 6 billion of us on this planet. So much for the "hair bridge" to Mars...

exabyte
1018 or 1 quintillion bytes. 260.

1,000 petabytes.

We could pretty comfortably fit everything ever broadcast or published within a few exabytes.

Clickery says 5 exabytes would hold all words ever spoken. The estimates for how many people have lived on Earth since the beginning are mostly around 100 billion. Using that number (1011), we can compute that 1018 words divided by 1011 people is 107, or ten million. So we have room in 5 petabytes for ten million words from every person who has ever lived on this planet.

Using a generous average life expectancy of 45 years (we didn't reach an average of 50 until after 1900), that's an average of 16436 days per life. Divide that into ten millions words and you find that each person is allotted 608 words per day. That sounds rather low considering some people I know. On the other hand, early humans were probably not great conversationalists, and time spent as babies will bring the average down, also. But if you're more comfortable using an average of 6000 words a day, we're still estimating on the order of exabytes (50) if someone had bothered to record all those words.

"exahair": Would require the hair of 10 trillion people, and if only 100 billion people have ever lived on Earth, you'll see we're in trouble now. That would be 100 haircuts for everyone who's ever lived. Side by side, 1018 hairs, each measing .05mm wide, would span 5x1013 meters. Neptune's average distance to the Sun is about 4.5x1012 meters. Double that to get the distance between the farthest reaches of our Solar System, and you get about 1x1013 meters. An "exahair" is five times that distance.

We're probably well beyond known physical limitations by now. But who knows what discoveries lurk in the future? If Moore's Law still continues to hold, that palm-size exabyte storage device would be ready by around 2040.

zettabyte
1021 or 1 sextillion bytes. 270. (Just counting this many bytes, we've finally exceeded the size of a 64-bit binary number, 264-1, for you computer geeks.)(And if you're reading this, you probably are...)(That means a 64-bit address space could address up to 16 exabytes? Yikes.)

1 zettabyte = 1 billion terabytes.

This is getting beyond comprehension.

With Moore's Law, we'd get there around 2055.

"zettahair": Side by side, this now stretches 5x1016 meters.

The speed of light is 186,287 miles per second. That's over 7 times around the Earth in one second. A light year is the distance light travels in one year. That distance is about 9.4x1015 meters.

Thus, a "zettahair" is more than 5 light years. The closest star to us, Alpha Centauri, is about 4 light years away.

yottabyte
1024 or 1 septillion bytes. 280.

That's a "yotta" bytes.

Moore's Law would have them for us by 2070.

Just in case you were wondering, a "yottahair" would be about 5,000 light years. (A "parsec" is 3.3 light years, so we could also say that a "yottahair" is just over 1500 parsecs. )

Our Milky Way galaxy is about 150,000 light years across. That's 30 yottahairs.

???
In case your curiosity overwhelms you, here are the terms for 1027, 1030, 1033, 1036,....

octillion, nonillion, decillion, undecillion, duodecillion, tredecillion,....

And if you can't keep all those zeroes straight, here's a page that let's you automatically convert your byte units

Back down to Earth
One "mole" of atoms is 6.022x1023atoms. That's more than "zetta" and less than "yotta". A mole is about 600 "zettas". 2 moles is a little more than a "yotta". There is no such term as zetta-atoms or yotta-atoms, but if there were, this would likely be the meaning.

One mole (6x1023) of H2O molecules has a mass of 18 grams. (16g of oxygen and 2g of hydrogen. The mass of each mole of atoms is its atomic weight in grams.) 1 milliliter = 1 cubic centimeter of water has a mass of 1 gram. So 18 cubic centimeters of water is 18 grams and thus is made up of 6x1023 water molecules. 30 cubic centimeters of water contains 1024 water molecules... or a "yotta" water molecules.

You know that really big number we were talking about a few paragraphs above? A yottabyte? Go sip 2 tablespoons of water. That's 30ml., or 1024 water molecules. A "yotta H2O".

* "Moore's Law" regards the exponential growth of chip density, or how many transistors fit on a computer chip. It's generally accepted as an estimate that the density will double every 18 months. The original prediction was made in 1965, and has held to be remarkably true over more than 30 years.

But beyond silicon, observers have noticed that the same "law" has held close to true in almost all aspects of computing: processor speed, memory, and mass storage. So while the "law" was not intended to apply to floppy disks, hard disk, CDs, and DVDs, it's prediction of doubling every 1-1/2 years has also held close to true for mass storage.

Ten "doubling cycles" would take 15 years by this "law". 210 = 1,024. Thus, every 15 years performance and capacity has increased times 1000, and we've moved up one notch on the name scale, from "kilo" to "mega" to "giga" and on.

** Roy Williams from Cal Tech created a comparison chart similar to this years ago. His original Internet link to it is now gone, but you can still find copies other people have posted by searching for his page title "Data Powers of Ten".

- Mark Pelczarski
Computer Science Department
Elgin Community College
graphicsmagician.com