How Heavy Is a Petabyte?

Jon Morgan writes "Whilst heaving around numerous data storage systems to sell (they weigh A LOT!), we got to wondering: How heavy is a Petabyte of data storage? Our best guess is 365KG, which is 6 million times lighter than in 1980! But is there a lighter way to store a Petabyte?"

Comments

library of congress

[SoupGuru]

How heavy is a Library of Congress?

Re:library of congress

[troutinator]

According the Library of Congress' website they have approximately 32 million books. A bit of googling turned up that an average book weight about 12 ounces. So, 32 million * 12 ounces = 10,886,216.9 kilograms

Re:library of congress

[RichardJenkins]

Asking a question like this is about as silly as asking how wide a year is. It's just not immediately obvious that this question makes no sense because it gets confused with the similar question 'what is the lightest device(s) capable of storing a petabyte of information.

Re:library of congress

[SoupGuru]

See, this is why I love slashdot. Ask a silly question and more often than not you'll get an answer.

Re:library of congress

[weirdcrashingnoises]

according to some website the LOC holds aprox. 10 terabytes worth of information.

which means that 102.4 LOC's would equal 1 petabyte.

10,886,216.9 * 102.4 = 1,114,748,610 kg

or aprox 2,457,600,000 lbs.

Re:library of congress

[camperdave]

A year is two AU wide, about 300 million km.

Re:library of congress

[Anonymous Coward]

I'm a Martian, you insensitive clod!

Re:library of congress

[Smidge204]

If we take a book to have approximately 7000 BTU per pound when incinerated (newsprint is about 7,500) then we get 437.5 BTU per ounce.

So 1 LoC = 14,000,000,000 BTU or 14,770 gigajoules.

Finally! A heat unit LoC equivalent!
=Smidge=

Re:library of congress

[W3bbo]

No, that's diameter of the earth's orbit around the sun. A year would be the circumference which is 940 million kilometers.

Re:library of congress

[Anonymous Coward]

10,886,216.9 kilograms or 10.9 kilotons is slightly less than one Hiroshima.

So if every book in the Library Of Congress was made of TNT and you detonated them all together, the total yield would be slightly less than one very small atomic bomb. Fuck.

Re:library of congress

[Ambiguous Coward]

So when some asks you "How wide is this circle?" do you tell them the circumference? If someone asks you, "How wide is this desk?" do you provide them the length of the perimeter?

I propose that your definition makes less sense than any of this. :)

Re:library of congress

[spyder-implee]

No, he asked how WIDE it was, not how long it was. So he was right in the 1st instance.

Re:library of congress

[Anonymous Coward]

A year is actually 0AU wide, 6 months would be 2AU

Re:library of congress

[chunkyq]

I love Slashdot. You, sir, have validated your existence for the day.

Re:library of congress

[W3bbo]

In my head I equated orbit length to "width of the year" since I was thinking of a timeline, and a timeline has width (from 1st Jan to Dec 31st) and at each point on the timeline the earth is at a certain point on its orbit which corresponds to a distance from its initial position.

Re:library of congress

[CarpetShark]

See, this is why I love slashdot. Ask a silly question and more often than not you'll get an answer.

And it's usually, "fuck off" ;)

Re:library of congress

[Cryacin]

What if we want a silly answer?

There's always dig.

Re:library of congress

[Anonymous Coward]

Thank God it takes your approval to validate his existence.

Re:library of congress

[t3h Odd one]

On that note, I suggest moving the White House or at least auction it off.

Re:library of congress

[nrlightfoot]

That's about half of what 1 terabyte of magnetic storage weighed in 1980, so I guess that in 1980 books still had better information density than magnetic media.

Re:library of congress

[dgatwood]

Which brings us right around to my solution for storing a petabyte. It only weighs a few pounds... on each end... of a very long distance. It involves three lasers with insanely precise tracking mirrors orbiting the sun at 0 degrees, 120 degrees, and 240 degrees around a circular orbit. This ensures that each laser can see both of the other lasers.

Modulate the beam with the data. If we naively assume one bit per Hz, and approximate it at 10^17 bits per petabyte, and if we modulate the beam at 10 THz, the total distance around the triangle has to be about 2 * 10^9 miles, or a little over 20 AU, putting their orbit a bit inside the orbit of Jupiter. The problems of how to actually track an object so precisely and how to modulate a laser at 10 THz are left as exercises for the reader. :-D

Re:library of congress

[mysidia]

Then ask congress. They'll spend a few hundred million $$$ researching the question.

Probably hire a team of researchers to go through the library of congress and painstakingly weigh each individual book, or a certain percentage of the books, for statistical analysis.

And then ammend the final report to say the books weigh, however much, the book lobbying groups want them to say they weigh.

Re:library of congress

[The_mad_linguist]

Well, we can convert distances into time intervals via relativity...

To nobody's surprise, the conversion factor is a well known physics constant. c.

So a year is exactly one light-year wide.

Re:library of congress

[rant64]

With that precise an answer, I think you might mean "exactly 32 million books" and an "average book weight of exactly 12 ounces". Personally, I think I'd just go with "11 million kilograms", unless you feel lucky. :-)

No! No! No! We'll end up with horrific rounding errors when somebody else picks up the number and tries to convert it into a measurement of angular momentum.

Re:library of congress

[shish]

digg isn't silly; they're entirely serious when they give answers that are life-endangeringly wrong :-S

Re:library of congress

[techiemikey]

unfortunately, the practicality of removing the BTU from a book is more effort than it's worth? Have you ever tried to burn a book? It's actually not as trivial as one would think.

Re:library of congress

[YourExperiment]

Okay. I am heavier than a duck, by a considerable margin.

Need conversion to units of Libraries of Congress

[AcidPenguin9873]

What are these Petabytes of which you speak? America measures data in units of Libraries of Congress.

MicroSD

[jeffb (2.718)]

...weighs something like 300mg/card. That's 48GB/gram, or a bit over 20g/TB, or 20Kg/PB.

Re:MicroSD

[Animaether]

But the smaller the chipsets, the larger - relatively - the packaging becomes. You can't just keep shrinking down the packaging, after all.. it would get far too flimsy.
So what you'd really need to weigh is the actual PCB with components, but sans all but a sliver of the bit that is the connector (the copper strips etched into the PCB to function as such).

Re:MicroSD

[Burning1]

Technically, if you don't count the hardware to read the data, we could simply remove the hard disk platters from the drive. Since most of the drive's weight is made up of the casing and read electronics, it would probably swing the data/weight ratio back in the favor of hard disks.

or 2.5" drives?

[whoever57]

I think that a 2.5 inch drive weighs less than half the weight of a 3.5 inch drive, so using twice as many of the 2.5" drives (available up to 1TB today) will reduce the weight.

Re:or 2.5" drives?

[Forge]

Problem is those methods of dropping the weight, also increase the cost (TFA assesses both).

My problem with the assessment however, becomes even more glaringly obvious when you look at the micro SD proposal in the grandparent. If you are going to have a single SD card reader and plug these cards in as needed, the weight estimate is ok. If however all 1 PB of data must be immediately available to your software, the weight gos up dramatically.

In the case of 3.5" SATA HDDs, that weight/cost should include a storage system that renders all the data available at the same time. 140 Lbs for 48 Hard drives is reasonable.

Depending on your RAID Level, 1,500 Lbs per petabyte is closer to reality. 1,700 Lbs to 2,000 Lbs per petabyte if you add the rack to the equation.

BTW: Doing something sane, like RAID, instead of JBOD or RAID 0, will increase that mass somewhat.

Re:or 2.5" drives?

[jeffb (2.718)]

Not if you do it reversibly.

There is a way!

[b4upoo]

It will take me a while but committing all that data to my memory won't add any measurable weight to me at all.

Re:There is a way!

[Palshife]

Right, but to answer the question, we still need to know your weight!

Re:There is a way!

[Thiez]

Insightful? Assuming you can perfectly remember 1 byte per second, you'd be memorizing for over 100 million years. The human brain is great and all that, but no way are you going to store that much data while being able to reproduce it later.

Re:There is a way!

[CorporateSuit]

Insightful? Assuming you can perfectly remember 1 byte per second, you'd be memorizing for over 100 million years. The human brain is great and all that, but no way are you going to store that much data while being able to reproduce it later.

Considering a single "frame" of vision for a pair of human eyes is estimated at 576 megapixels (truncating at peripheral vision). We'll imagine that each pixel is assigned a 16-bit hexadecimal value. That means, each time you glance at something, each frame would be calculated at a little more than 1/1000th of a terabyte. The lowball framerate for the human eye is about 18 frames/second (things look fluid). That means that every 50 seconds, your eye is downloading a terabyte of information. He'll absorb it in less than a day through eyesight alone. That doesn't include audio, olfactory, touch, or taste. His brain's data compression will downsize a lot of that information, so it will take him more than a day, but for your i/o ports, taking in a petabyte of information is a daily task.

You'd be hard-pressed to find a living organism that downloads information at 1B/sec

Re:There is a way!

[flappinbooger]

I think there was a movie about this type of thing.... But I don't remember what it was called...

Johnny something-or-other....

Re:There is a way!

[The_Duck271]

It is interesting to speculate about the information density of memories. Petabytes may impinge on our retinas but much less will make it to the brain and only a tiny fraction will make it into long-term storage: my memories, at least, are not 20,000x20,000 pixel video. They're more like crude reconstructions of small fragments of audio or video. As if the original data had been brutally compressed, then uncompressed and filled in by an artist applying a lot of guesswork. I would guess that we store surprisingly few bytes for each megabyte of input.

Re:There is a way!

[greyhueofdoubt]

Pointless. You could put 1 billion people on a scale and have them all memorize data as fast as they could and the weight won't change (correcting for evaporation, etc.).

-b

Re:There is a way!

[HeadlessNotAHorseman]

>>Insightful? Assuming you can perfectly remember 1 byte per second, you'd be memorizing for over 100 million years.
>>The human brain is great and all that, but no way are you going to store that much data while being able to reproduce it later.

Actually, it would not be quite that difficult if the data consisted of pictures. If we take the IA-60 definition of Word as 8 bytes. And a picture is worth 1000 words. So that's a total of about 137,438,954 images to memorise, which at a rate of 1 per second would take 4.4 years.

Cloud computing

[Sta7ic]

Just stick the petabyte on the cloud! Clouds are as light as air!

(why yes, I am from Marketing, why do you ask?)

lim-0

[mattj452]

Since data storage is just one case of transmission channel (just sending it through time, not space) you can store the 6 Petabytes in a transmission. All you need to do is place one sender here, and one eh, let's say at the end of the Universe. As long as the data is being transmitted, it doesn't really weight anything. Yes silly question will get a silly answer :)

Re:Need conversion to units of Libraries of Congre

[luton]

I've seen stats that all the books ever written by mankind add up to 50 PB of data storage. Presumable unZipped :)

but-electrons-don't-weigh-anything

[John Hasler]

Whatever gave you that idea?

About 2 Kilos

[BBCWatcher]

Nobody knows exactly how much data the average human brain can hold, but one estimate is 500 to 1000 TB. If the average adult human brain weighs about 1.3 or 1.4 Kilos, then "about 2 Kilos" would hold 1 Petabyte.

Re:About 2 Kilos

[Kaeles]

I'm no expert in this field but I think the link that you provided had underestimated the human brain by many orders of magnitude. The human brain is not a hard drive. I don't think there is even any counterpart to it in current computer technology (maybe quantum computing?), whatever that is, so the comparison is meaningless. The brain doesn't just "store" information like a hard drive. It analyses, modifies, categorises, correlates, extrapolates, fills in missing blanks, filters and blanks out others and many other things that we are just beginning to discover. For example, a human child will quickly grasp the concept of doors and doorknobs, without any "programming" (I've had toddlers so believe me on this). This is why I think A.I. enthusiasts will ultimately fail.

People like you drive me nutters. The human brain has billions of years of evolutionary programming built into the seperate layers of the brain, there are so many built in functions that we don't even realize it in normal everyday activities. For example, your brain is "hardwired" from birth to recognize human faces, and to emit "happy juice" when the faces are familar or matched with motherly smells. Just because its not programmed after birth, does not mean that the hardware itself is not built for the task. This is no different from creating a custom asic or fpga for doing GA's or ANN's.

Re:About 2 Kilos

[Taddy Tadbag]

For example, your brain is "hardwired" from birth to recognize human faces, and to emit "happy juice" when the faces are familar or matched with motherly smells.

My 'happy juice' can be stimulated to emission by data from a storage device that only needs to be able to store a few hundred KB in JPEG format.

"But is there a lighter way to store a Petabyte?"

[John Hasler]

Sure. Store it in a WOM chip. They only weigh a few grams, hold literally unlimited data, and are really fast.

Re:Need conversion to units of Libraries of Congre

[HotNeedleOfInquiry]

And after you decompress it, you get 48 and a buttload of fragmented chains.

Minimum mass of a Petabyte

[Chris Burke]

Thinking about the decrease in mass of a petabyte got me thinking about Information Theory and the minimum energy required to store a bit. Or rather, to irreversibly manipulate one bit of information, which I think describes the act of writing to any kind of RAM (disk or otherwise). If I extrapolate that to also mean a mass whose rest energy is sufficient to manipulate a bit, that could give the theoretical minimum mass for a bit of storage. I don't actually know enough information theory to know that value, or even if the comparison from energy of information manipulation to mass of storage is valid, but it struck me as interesting and maybe somebody knows? What's the minimum mass of a petabyte?

Re:Minimum mass of a Petabyte

[the_other_chewey]

For example, it seems to me like a "full" drive seems to physically weigh more than a blank one, sort of like a full battery is noticeably heavier than an empty one.

Wrong on both counts. A "full" magnetic hard drive platter just has its magnetic domains aligned in a certain pattern.
Those domains are physically there whether they are used for data storage or not. So the weight will be indentical.

A battery does indeed become lighter when "emptied" - according to E = mc^2 and the energy that came out of it.
However, this is way, way, way under anything you would be able to notice.

An AA alkaline battery can deliver about 10000 Joules (http://www.allaboutbatteries.com/Energy-tables.html) - so
a discharged (= "empty") AA alkaline will weigh m = E/c^2 or roughly 10^-10 grams less than a charged one.

That's 0.1 nanograms. About 100 human skin cells. No, you won't notice that.

Re:Minimum mass of a Petabyte

[monopole]

That was my dissertation topic, conventional systems require ~kT per bit (k is the Boltzmann constant = 1.3806503 Ã-- 10-23 m2 kg s-2 K-1 and T is the temperature of the gate in Kelvin) for each read. Quantum systems can access well below that by various trickery (single photon optical computers can reduce this by a thousandfold). In theory a individual photon can hold huge amounts of data in it's state vector before collapse. The trick is making a measurement on enough of these photons to extract the info you need while overcoming shot noise.

Re:Work it out in your head

[John Hasler]

They probably want an error rate lower than 10%.

It depends..

[Qwell]

Are you storing mostly 1s or mostly 0s? Everybody knows they don't weigh the same.

Re:It depends..

[IHawkMike]

Here's an interesting discussion on the topic ;)

Re:Theoretically quite close to zero ...

[RajivSLK]

Or you could just stick a mirror "out there". The light would quite conveniently come back at you. Or you could sneak around the other side of the universe and wait for the light...

A lot heavier than...

[marcus]

and a lot bulkier than...

a few strands of DNA.

Re:A lot heavier than...

[Anonymous Coward]

and a lot bulkier than...

a few strands of DNA.

you would have to twist words to make your point!

Re:A lot heavier than...

[Artraze]

Well, a rough check shows that each base pair (and backbone) weighs about 614amu, which gives a weight of 10^-21 grams for 2 bits. So, pure DNA weighs about a 4ug per petabyte, supposing my calculations are correct.

However, that's hardly fair. The density of bits is _far_ from the density of the actual storage. After all, a hard disk uses only extremely small regions (probably only a few million amu) on the surface of a disk. However, the motors, the case, and even the disk (substrate) itself are orders of magnitude heavier than the bits themselves. I'd be rather surprised if the actual storage was much more than a couple grams.

The point is, of course, that there are all kinds of ways to store data, but when it comes down to weight, the control mechanisms are what matters. For this reason it's extremely unlikely that DNA will _ever_ be used as storage, except if we start making bio-computers.

Also, for what it's worth, the human genome only stores about 770MB, only a bit more than a CD.

Re:How much does a "full" HDD weigh vs. an empty H

[Darkness404]

Oh, and BTW, when a person dies does the body weigh a tiny amount less after the sole leaves?

Depends on the shoe they are wearing. On a boot, no, its a large amount, on sneakers, yes it might be a tiny amount.

Re:Cloud computing-Clouds in Elephant Units

[Nom du Keyboard]

Clouds are as light as air!

A common misconception, and just saying it on Slashdot doesn't make it true. Clouds weigh more than elephants - much more. In fact, you can learn the weight of clouds in elephant units here.

Not only that, but clouds are usually darker than the air around them.

Already answered

[slasho81]

This subject has already been discussed.

Try using Micro SD cards instead

[kroyd]

With 32gb cards weighting 0.5 grams one terabyte should require 32 cards, or 16 grams. 1024 terabytes should then weight 16384 grams, or a bit more than 16kg.

I don't think there is a storage media with higher density available commercially right now - and probably not until the 64GB microsd cards becomes available.

Re:Cloud computing-Clouds in Elephant Units

[imsabbel]

Air also weights more than elephants.
In fact, every square meter of the world has 2 elephants of air on top of them.

So "missconception" my ass.

Re:Mass!=Weight

[nsayer]

Since they started making scales calibrated in kg instead of Newtons.

Re:How much does a "full" HDD weigh vs. an empty H

[Alyred]

Yeah, but that's called "voiding their bladder" or the even more unpleasant related process.

Re:Cloud computing-Clouds in Elephant Units

[geekoid]

Being pedantic is the wrong place to fail , like you did.

You failed to take the weight of air into account. Why, when you do that they are, in fact, lighter then air.
Otherwise they would fall down, and we call that 'rain'

Re:Cloud computing-Clouds in Elephant Units

[clarkkent09]

Thanks for nothing Peggy LeMone. Saying that a "typical cumulus cloud" weighs as much as 100 elephants is a meaningless statement without giving us a hint as to what the hell is the size of a "typical cumulus cloud". I bet we are talking about a hell of a lot greater volume than 100 elephants.

Here is an equally interesting piece of news for you: a pile of feathers is heavier than a tank! Of course it all depends on the size of the pile but who can bother with those little details.

Re:Need conversion to units of Libraries of Congre

[syousef]

I've seen stats that all the books ever written by mankind add up to 50 PB of data storage. Presumable unZipped :)

You've seen ESTIMATES.

Re:Cloud computing-Clouds in Elephant Units

[Anonymous Coward]

A common misconception. Weight is not mass.

Air also has mass, not much different from the mass of cloud (which is mostly air).

Re:Need conversion to units of Libraries of Congre

[ducomputergeek]

51.2 LoC's

Assuming LoC is still = 20TB

Re:Cloud computing-Clouds in Elephant Units

[Nom du Keyboard]

In fact, every square meter of the world has 2 elephants of air on top of them.

Tell that to my doctor the next time I stand on his scales.

Re:Work it out in your head

[nacturation]

Well, with the right RAID (Redundant Array of IDiots) scheme, the human brain could be harvested for perfect storage.

Re:Cloud computing-Clouds in Elephant Units

[rjhubs]

and all of this sits upon a tortoise? Amazing!

Re:Work it out in your head

[MichaelSmith]

Well, with the right RAID (Redundant Array of IDiots) scheme, the human brain could be harvested for perfect storage.

Finally I understand why /. exists.

Re:Cloud computing-Clouds in Elephant Units

[Tanman]

Here's the one that'll really get you:

A pound of feathers weighs more than a pound of gold.

Re:No, a bettery wouldn't get any lighter

[dido]

But you are converting mass into energy and energy into mass even in this case, although the amounts are ridiculously small in the case of chemical reactions, which is why conservation of mass is a more than reasonable approximation in chemistry. The mass is stored in the molecular binding energy of the battery's chemicals, and converted into the energy used when the battery discharges. For example, if you weighed very very carefully a bunch of hydrogen gas, a bunch of oxygen gas, and the water you got after combining the two (in a fuel cell reaction, which we can think of as the simplest sort of battery from a chemistry point of view), the water would weigh ever slightly less than the hydrogen and the oxygen, though the difference would be extremely small, since the binding energy difference of a water molecule versus that of hydrogen and oxygen molecules is only a few tens of electron volts, about 10^-35 kg or thereabouts, which amounts to a difference of about a quadrillionth of a gram for one mole of water. For nuclear reactions though, the binding energies we deal with are millions of times greater, and E=mc^2 is much more obvious. For instance, in the nuclear fusion of the two helium-3 nuclei to produce one helium-4 and two free protons, the helium-4 and the two protons weigh less than the original helium-3 nuclei by about 12.86 MeV/c^2, or about 6 milligrams less than if we started with a mole of helium-3 at the beginning of the fusion reaction.

Re:Cloud computing-Clouds in Elephant Units

[DMUTPeregrine]

That is, of course, due to the fact that a Troy pound (used for gold) measures a lower weight than an avoirdupois pound.

Re:Cloud computing-Clouds in Elephant Units

[cerberusss]

No, not just one. It's tortoises all the way down, young man.

Re:Need conversion to units of Libraries of Congre

[Philip K Dickhead]

Exactly. Once you find out, the cat dies.