How much is a Petabyte? | Hacker News Reader

How much is a Petabyte?

mozy.com

32 pointsquizbiz24 comments

Kinda cool, but the second one (with the concentric circles) is obviously a bad graph. The radii of the circles are proportional, so the ratios of the areas (apparent size) are off by a factor of pi.

Pi doesn't affects the ratio. However - the ratio of the areas is the square of the ratio of the radiuses, and that is the mistake in the graph.

(Sigh), yeah so in this case they were off by a factor of 3, not pi. I was close :)

Sorry for being grumpy, but how does it help to say that "1 petabyte = 20 million four-drawer filling cabinets filled with text"? Changing one huge number (1pb) with another (20M) doesn't add much to the overall understandability. The other examples are similar in nature. They're either comparisons with other giant magnitudes (10B facebook photos) or comparisons with things we have no fucking clue of how big they really are (HD manufactured in 1995, google's processed data).

"1 petabyte is a lot of data". Oh rly?

It appeals to our experience with everyday things. We have a mental model for how much information is on a piece of paper, in a book, or in a second of video.

I don't think anyone can properly visualize 20 million of anything.

No, we can't. But "20 million cabinets of text" has more meaning to us than 10^15 bytes.

That's what - one Empire State Building worth of cabinets ?

100 floors * 100 offices/floor * 20 cabinets/office = 200k cabinets ...

Hmm. Off by a couple of orders of magnitude. More buildings or squeeze until the pips squeak.

Seems the E.S.B. volume is about 1 million cubic meters and a 4 drawer cabinet is about 0.4 cubic meters. So, ten buildings filled solid.

Incidentally, data via Google. Wolfram Alfa pouted on the building query.

(Yes, I'm easily amused. Or distracted: http://xkcd.com/356/)

I had the same feeling. It doesn't help you fathom anything at all.

Looks like they forgot to show it to a geek before publishing. Not only did they claim a TB is 1024 GB, but they did so just after saying a TB is 1000 GB.

Also: interesting to note that the graph of dollars per GB has continued to decrease rapidly. Consumer-level drives are currently at about $0.08 per GB.

I don't see that anywhere except for the part where they talk about the hard drive holding 1000GB. That's the standard way of measuring hard drive space for most drive manufacturers.

About halfway through they state 1 TB = 1000 GB, which is correct.

Further down, they have comparisons of relative size from MB to PB, where they incorrectly use binary units.

It is correct to use binary units unless you are selling hardware. We say TeraByte, but we mean TibiByte.

The 1998 price seems way off. IIRC I had a 4GB hard drive that year in a computer that cost at most $2000. I doubt more than half its cost came from the hard drive.

I'm calling bullshit on the last stat. Where does he source this from?

I agree. This is ignoring the fact that there are hundreds of languages we don't know about and probably millions of texts unaccounted for lost through the ages.

... which are all probably not relevant in scale.

It's like the "how many humans ever lived?" question. People immediately start discussing about the definition of a "human being", homo whatever. I say it's pretty much irrelevant to the final number.

I agree, initially for a different reason to the other commenters here - text doesn't take up that much room.

If you assume that each book is translated in each language then the figure is way short. But it's probably better to interpret it as each book in its original language.

I've read a figure of 7000 languages (http://en.wikipedia.org/wiki/Ethnologue). A survey discussed here (http://answers.yahoo.com/question/index?qid=1005120803482) suggests up to 175 million books ever published. That's 1.2 trillion uniques.

But if we just worked with the 175 million figure then 50 petabytes would allow 300 megabytes per book. In fact assuming say 3 megabytes (base 2, mibibytes) per book you could fit just under 18 billion books.

My figures are a little off from just using base 2 instead of base 10 but you get the gist of these back of the napkin numbers.

Probably ∫ estimated human population · estimated literacy rate · estimated hours per year a literate person would spend writing · words per hour · dt, or something like that. Given the dramatically higher levels of population, literacy, leisure time, and typing speed in recent decades, you should be able to get a pretty good estimate that way.

I don't feel like a petabyte is a lot of data anymore.

Sun's Thumper is a 4RU cabinet containing 48 disks. If you had a rack full (10 thumpers: 40 rack units) with 2TB disks, that is 1 raw/unformatted petabyte.

Indeed. I still remember when buying a 3GB harddisk was a big deal and thought that you could never fill it. And now I need more than 300GB just for my laptop.

Now you can get a 1TB one for less than a $100.

A lot of dogs and cats

Your whole life recorded on video.

That much.

Reply on news.ycombinator.com