Peter Chubb posted a patch to the lkml, with which he’s now managed to mount a 15 terabyte file (using JFS and the loopback device). Without the patch, Peter explains, “Linux is limited to 2TB filesystems even on 64-bit systems, because there are various places where the block offset on disc are assigned to unsigned or int 32-bit variables.”
Peter works on the Gelato project in Australia. His efforts include cleaning up Linux’s large filesystem support, removing 32-bit filesystem limitations. When I asked him about the new 64-bit filesystem limits, he offered a comprehensive answer and this interesting link. The full thread can be found here on KernelTrap.
Reaching beyond terrabytes, beyond pentabytes, on into exabytes. I feel this sudden discontent with my meager 60 gigabyte hard drive…
Interesting, as my trusty BEOS bible lists 1000 terrabytes as a petabyte, not pentabytes (p299)
However, several sites on the internet do list pentabytes, maybe that is the name for 1024 terrabytes, and not 1000?
1024 and 1000 have differant prefixes, i forget what they are for 1024 and such, something like gigibyte and such. It’s just the problem of the bases, the sane world thinks in base ten, computers in base 16 (well in storage). Just be glad there isn’t a English/Empirial/SAE system of size for computers.
Where is this sane world you speak of? I can count to 31 on one hand so i either have a bizare number of fingers, or a better way of doing things.
I bet ya if we had 8 fingers per hand we would prefer
base 16 over base 10
No, computers are in base 2, even storage (0 and 1).
…like… to store next decade’s Microsoft operating system and applications… no room for data…
They should start bloating the user-data files to justify hard drive sizes too…
(yes, I know people doing audio and video need as much storage as they can get, I’m just being silly)
Peter Chubb posted a patch to the lkml, with which he’s now managed to mount a 15 terabyte file
Man, I wish I had that hard drive
“penta” does not exist. Nor “terra”
The correct names are:
10^12 = “tera” (T)
10^15 = “peta” (P)
Then there are the IEC 60027-2 standard prefixes for base-2 numbers:
2^40 = “tebi” (Ti)
2^50 = “pebi” (Pi)
FAT-16 was initially limited to 32MB (16-bit times 512 bytes). Instead of inventing FAT-32 right away, MS invented the clusters, allowing FAT-16 to grow all the way to 4GB. If you just so the same trick on linux, you can go all the way to 256TB with the same factor.
Sure, the idea works much better for a FAT than for a unix system (because FAT doesn’t have inodes), but it can be a good enough temporary solution that doesn’t necessarily have the risk associated with a major kernel change where you have a small risk of missing one variable somewhere(admittedly, goot test frameworks cen be developed – and 2TB isn’t that hard to reach, even with phisical disks – a single SCSI channel fully populated with 181GB drives will give you 2.5TB).
JBQ
We (almost) all know that leap years occur every 4 years, except every 100 years, except every 400 years.
Did you know, however, that a much precise approximation would have been to space them every 4 years except every 128 years?
I know, I know, useless tidbit…
JBQ
Fooks, yep we would, course base 16 would be fine if we came up with new numbers having to switch to letters part way through things would be better.
16 == g is just to confusing
I relize computers are base 2, but base 2 to hexidecimal and back is quite easy. I think we will get real confused if they ever get turnary and up computers working.
Pentabyte is a real word.
one pentabyte = 1024 terabytes
1024 not 1000.
This news story came from Jeremy Andrews and published word by word as received. Your complaints to Jeremy please, as I got no idea how you call these words in english.
Elmer: No, 1024 terabytes is a petabyte.
As someone wrote on slashdot, the only thing a “pentabyte” could be is FIVE bytes. 🙂
Sorry, but “penta” is not an official prefix for units. Not decimal nor binary. RTFStandards!
Hey JBQ, you fell for the classic slashdot newbie mistake. Remember to read the linked article in full first
The architectural problem is above the filesystems, in the block layer. Changing the block size for all disk blocks is a much more horrible and fundamental change than this patch from Peter.
The individual file systems are mostly unaffected. XFS, JFS and Reiser were all designed to handle > 2TB filesystems and will work (modulo implementation errors) once Peter’s fix is applied. Most older filesystems stop before 2TB anyway for one practical reason or another, and ext3 needs a userspace tweak to allow > 2TB filesystems to be built but is otherwise fine for a few more terabytes yet.
None of this affects 32-bit Linux systems, the limit there is still 2TB per filesystem. I can’t believe we still have to worry about this. Shareholders ought to be screaming for blood at Intel.
Pentabyte is a real word.
One pentabyte is 1024 terabytes.
I don’t care if some CS proff tried to change the names, pentabyte is real.
http://translate.google.com/translate?hl=en&sl=de&u=http://translate.google.com/translate?hl=en&sl=de&u=http://home.t-online.de/home/burmi/index_frameaufbau.html?helferlei…
Well, it seems that there are different oppinions about that. Here are a few links stating that a petabyte is 1024 (or 1000) tera bytes.
But this shouldnt really be an issue in the first place. Go look up the units and it should be clear. I list all relavant units here for your convenience
Penta = 5
Kilo = 1 * 10^3
Mega = 1 * 10^6
Giga = 1 * 10^9
Tera = 1 * 10^12
Peta = 1 * 10^15
Exa = 1 * 10^18
And according to the last link it continues with
Zetta = 1 * 10^21
Yotta = 1 * 10^24
Three quick links explaining what a petabyte is:
http://www.webopedia.com/TERM/P/petabyte.html
http://searchstorage.techtarget.com/sDefinition/0,,sid5_gci212777,0…
http://content.techweb.com/wire/story/TWB20010226S0026
A fun link also calling it a petabyte:
http://www.ccsf.caltech.edu/~roy/dataquan/
And finally, try doing a google search for both words.
The result was:
Pentabyte: 102 matches
Petabyte: 10800 matches
As for the different names for using multiplies of 1000 or 1024, they carry the same name, which might be confusing, but it really is the modulo 1024 names that abuse the naming convension, however there are no standard name to replace them*.
So quit this talk about pentabytes, the meaning of penta is 5, and only 5. If anyone tries to say differently then they are wrong
*) That is, there are no actual recognized standard name to replace them. I did read about the gigi, etc names, but those hardly qualify as a standard.