How much physical space would an exabyte of data require?

[Flibble]

Your DNA already has the ability to scrub out transcription errors and mutations, I guess you want an enhanced version though?

[RocketSquid]

It is possible to make more stable alternatives to DNA with different materials in their spines, using DNA as an intermediate step. However, most of the study has been into their stability against heat or against biological/chemical attack, not against irradiation, so I'm not sure if they're actually more useful. 

[kerbiloid]

DNA → radiation → replication

Let the DNA data self-replicate.

Say, ~200 PB/g

According to
https://www.quora.com/What-is-the-total-mass-of-DNA-in-the-average-65kg-human-body
there is 6..60 g of DNA / 65 kg human body = 1..10 g / 10 kg.

According to
https://med-books.info/venericheskie-bolezni-kojnyie/obschaya-harakteristika-gribov.html
fungi cells contain 2.5..10 times more DNA than bacterial cells, but 200..300 times less than plant and animal cells.

We don't want the HDD running around the ship or requiring bees to create backups.
We want a think tank.

So, a fungus looks appropriate.

200 PB/g * 10 g/kg * (1/200) fungus/animal ~= 10 PB/kg.

Say, keeping 10 backups at once ~= 1 PB/kg.

So, a large barrel of fungus looks like a ship brain.

[kBob]

Very little once we have electron spin based memory.

[wumpus]

5 hours ago, kBob said:

Very little once we have electron spin based memory.

I'd have to wonder just how large the device to induce and measure the spin of individual electrons would be.  Hard drives gradually shrunk in size from dishwasher sized (when initially developed, storing I believe 30MB on a *lot* of platters) to the 5.25" drive of the 1980s and 3.5" drive from the 90s- (in desktops and now typically servers) and the 2.5" (slightly later, but only in mobile).  There was an attempt by IBM to make a hard drive roughly the size of a [US] quarter, it may have shipped in some Ipods but was quickly outpaced by flash.  Optical media has stayed constant since reducing in size from laser disc [analog video] to CD.

Flash storage doesn't appear to have size limitations.  As far as I know, even microSDXC has extra size for humans to have something to hold and manually insert.

I wouldn't be surprised if electron spin based storage has hard limitations in size, but might be directly done by a chip (and who knows, possibly a "nanomachine" rotating plate).

Also the "memory"/"storage" dichotomy can get a little rough at the edges: memory needs to be accessed in groups of 512 bits (one 64-byte cacheline) while storage needs to be accessed in 32kbit groups (one 4k page).  Latency also makes a big difference between the two, and I'd expect that dealing with individual electron spins would be so noisy as to require something more like what modern systems use with the 32kbit arrays: LDPC or similar iterated error correction schemes that require at least considerable nanoseconds of extra delay.  For a real world example, Intel is trying to sandwich their (well theirs and Micron's) new 3dXP memory as being able to be used for either (and selling the "512bit DDR-like interface" this year).

[kBob]

2 hours ago, wumpus said:

I'd have to wonder just how large the device to induce and measure the spin of individual electrons would be.  Hard drives gradually shrunk in size from dishwasher sized (when initially developed, storing I believe 30MB on a *lot* of platters) to the 5.25" drive of the 1980s and 3.5" drive from the 90s- (in desktops and now typically servers) and the 2.5" (slightly later, but only in mobile).  There was an attempt by IBM to make a hard drive roughly the size of a [US] quarter, it may have shipped in some Ipods but was quickly outpaced by flash.  Optical media has stayed constant since reducing in size from laser disc [analog video] to CD.

Flash storage doesn't appear to have size limitations.  As far as I know, even microSDXC has extra size for humans to have something to hold and manually insert.

I wouldn't be surprised if electron spin based storage has hard limitations in size, but might be directly done by a chip (and who knows, possibly a "nanomachine" rotating plate).

Also the "memory"/"storage" dichotomy can get a little rough at the edges: memory needs to be accessed in groups of 512 bits (one 64-byte cacheline) while storage needs to be accessed in 32kbit groups (one 4k page).  Latency also makes a big difference between the two, and I'd expect that dealing with individual electron spins would be so noisy as to require something more like what modern systems use with the 32kbit arrays: LDPC or similar iterated error correction schemes that require at least considerable nanoseconds of extra delay.  For a real world example, Intel is trying to sandwich their (well theirs and Micron's) new 3dXP memory as being able to be used for either (and selling the "512bit DDR-like interface" this year).

Well like all promised tech we will have to wait and see if it becomes practical.  In theory the density can go very high which is why it's being worked on.  So it might have some min size but for large storage requirements it would be the best.  But then I'm still waiting for my fussion generator and the cure for baldness .

First HD I worked with was 5MB about the size of a toy poodle.

[Kerbart]

On 5/13/2019 at 7:52 AM, Lukaszenko said:

It did cross my mind, but it can't be that hard to shield 3 cm^3. 

What unit is that? Or do you mean cm³?

[kerbiloid]

Has anybody mentioned Planck length?

~1.6*10^-35 m.

So, you can't resolve more than 1/(1.6*10^-35 )³ ~= 2*10¹⁰⁴ bits per m³.

 

Total internet data ~= 1.2 exabytes ~= 1.2*10¹⁸ bytes ~= 10¹⁹ bits.

Having 1000 redundancy, it requires ~= (1.2*10¹⁹ * 1000 / 2*10¹⁰⁴ ) ^1/3 ~= 4*10^-28 m wide cube of pure space to store.
 

Edited May 25, 2019 by kerbiloid

[Lukaszenko]

On 5/25/2019 at 8:16 PM, Kerbart said:

What unit is that? Or do you mean cm³?

Yes, what you said.