RHIC Computing Facility Crosses the 1 PB Mark

Martin writes "Brookhaven National Lab's RHIC Computing Facility (RCF) announced yesterday that the amount of data from the physics experiments at the Relativistic Heavy Ion Collider (RHIC) crossed the 1 PetaByte mark. A mail that was sent around to the RCF users contained a GUI screen shot (which is removed from the mail archive) that showed the number of MegaBytes transferred as 1,000,400,143. The RCF web pages have some pictures of the tape silos that hold the data. RHIC and the experiments have been discussed on ./ a few times, look here, here, and here."

Comments

Re:Just how big is a petabyte...

[Idarubicin]

Avogadro's number is approx 600,000,000,000,000,000,000,000 atoms per mol. If you were to store an an exabyte of data in one mol of material then each byte would have a budget of about 600,000 atoms. That may be doable...

It's eminently doable. That's 75,000 particles (atoms or molecules, depending on the species used) per bit--a huge number, still.

The problem comes in storage and readout. If I have to flip bits manually using a scanning electron microscope, that's no good.

On the other hand, let's assume that the work can be done optically, using a scanning laser. Take something the size of a vitamin E molecule; it absorbs visible light readily. Lying flat on a substrate, it would have a surface area (*very* roughly) of about 75 square angstroms. 75,000 of those would cover an area of about five million square angstroms. If arrayed over a square, that's about 240 nanometers on a side, or the diffraction limited spot size of a 480 nm wavelength laser.

Yep, it could be done. A monomolecular layer on a flat substrate; about half a kilogram of molecule. Perfect--a petabyte for your laptop! But--that would cover a total square area of six or so square kilometers...somewhat awkward to scan with a single laser, and a bit clumsy to carry.

Re:Just how big is a petabyte...

[Idarubicin]

I think we may have already acheived and surpassed this kind of information density. Does anybody know how much mass is represented by the media layer on a hard drive?

Off the top of my head, I don't know the mass of the media layer on a hard drive, but it's easy enough to find the areal density of the data.

IBM's 'pixie dust' (AFC) technology promises densities of up to about 100 billion bits per square inch. That's (roughly) 1000 square angstroms per bit. In other words, about an order of magnitude more area than in the grandparent post's fanciful thought experiment. Not only that, but the layer of iron oxide used in hard drives--though very, very thin--is not monomolecular.

Note that the great grandparent poster was asking about an exabyte of storage--1000 petabytes, or a million terabytes. Shaving three orders of magnitude off of that makes the problem much more manageable--though RHIC has still done a heck of a job accumulating a petabyte of storage space.