The Ultimate All-In-One Storage Solution 387
karnifex writes "Filled up your LaCie Bigger Disk already, and looking for a little more storage space? Good news! The Petabox is ready! 'The petabox by the Internet Archive is a machine designed to safely store and process one petabyte of information (a petabyte is a million gigabytes).' And luckily, as the Internet Archive notes, it's shipping-container friendly (20' x 8' x 8'). So save on delivery costs and order two!"
Petabox is ready! (Score:5, Informative)
From the article:
PILOT STATUS 5/2004
* The first 100TB Rack is up and running!
* The second 100TB Rack will be up by the end of May
Apparently this is some new use of the word "ready" with which I am not familiar. Neat technology, no doubt, but it doesn't really look like it's ready for prime time just yet.
Don't get too excited (Score:4, Informative)
Re:wrong (Score:2, Informative)
Re:"a million gigabytes"... (Score:3, Informative)
As always, wikipedia [wikipedia.org] has the answer(s):
Damn! Ambiguity!
Re:Price? (Score:5, Informative)
Rack materials cost is currently estimated to be $121K for 96TB. Node materials are a just under $1450. This price does not include markup, assembly or burn-in from the system integrator and thus will increase by another 5-7% to approximately $130K/rack.
The weight of a fully-loaded rack is estimated to be 1500 lbs. That figure may rise depending on what hardware is required for rack cooling.
Power is estimated to be 5500 watts. This too will depend on rack level cooling equipment.
These figures assume no external 1G Ethernet NICs.
For a breakdown of all the above, see the attached spreadsheet.
The cost... (Score:2, Informative)
http://www.archive.org/iathreads/post-view.php?id
Re:Useless Statistics! (Score:5, Informative)
50
How many 128kbps MP3s can you store on it.
250-300 million depending on song length
And most importantly, how many floppy disks is this equivalent too?!
700 Million - nearly 40,000 miles when laid end on end, or about 1500 miles when stacked on top of each other.
Re:Useless Statistics! (Score:5, Informative)
Glad you asked. Assuming that we have a 10^15 byte disk (which is how those decimal-loving hard disk manufacturers would define it), and your MP3s are encoded at 128kbps (where 1 kb = 1024 b = 128 B, as the binary folks would have it), then you could listen to MP3s nonstop for:
10^15B/(128kb/s * 128B/kb) = 61035156250 seconds
Cheers,
IT
Re:colossal... (Score:5, Informative)
The current genome [nih.gov] build has a size of 3,020,300,000 bp, at 2 bits per bp and 5(?) spice girls, that's about 3.5 GB (uncompressed).
Of course with a mostly static database like that you only want to store the diffs, not the whole thing. The bulk of the diff would be SNPs, roughly 1 per 1000 bp: 3,020,300,000 / 1000 / 4 / 1048576 that's about 0.72MB per spice girl. An if you only store the ones actually different from wildtype you probably don't need more than 20% of that.
You can fit a Spice Girl on a floppy.
Re:Price? (Score:3, Informative)
From the forum [archive.org]:
Rack materials cost is currently estimated to be $121K for 96TB. Node materials are a just under $1450. This price does not include markup, assembly or burn-in from the system integrator and thus will increase by another 5-7% to approximately $130K/rack.
So, about $1.3M (10 racks)
cLive ;-)
Re:To give you an idea of how much that truly is: (Score:5, Informative)
Re:Useless Statistics! (Score:4, Informative)
> 50
According to this article, a Library of Congress is approximately 10 TB (who knew--this obtuse metric actually has a measurement!!!)
http://articles.findarticles.com/p/articles/mi_
So the device actually can contain 100 Libraries of Congress.
Re:Business idea (Score:2, Informative)
It's called a Fibre Channel controller. Fibre Channel loop (which disks use) offers a total of 255 addresses - which has to include the controller. Disks now available in the 300Gbyte region, so 80 Tbyte/loop seems reasonable (and, according to the article, they seem only to have 100Tbyte up so far). 12 of these loops will give you your petabyte. Mind you, you will waste the disk bandwidth; this will gicve you capacity but not throughput.
Re:wrong (Score:3, Informative)
Actually the SI defines the prefixes irrelevant of units used. Think of the mil ('milli-inch'); how many do you think there are in the inch? If I had a thousand cats I could refer to the set as one kilocat, and hence if I had 1024 cats I could refer to it as a kibicat, Tweety-pie style; note that a cat is not an SI metrological term. Try playing around with the units(1) command sometime; to get a feel for these SI prefixes.