WD thrusts forth its mighty 3TB internal hardness Western Digital has unveiled the world's first 3TB internal drive, a Caviar Green model with four platters. This is a vast, astounding capacity, considering many desktop computer systems are operating with less than a terabyte of drive capacity. Seagate introduced its FreeAgent GoFlex 3TB drive in the summer and that was a …
but does anyone know what the real-life (not claimed) reliability of such disks is? I worry that we're trading headline capacity (mainly used to store movies you're never going to watch again) for long-term reliability that we can trust. We're already being taught that we should use RAID for ultimate security (so long as the controller doesn't go wrong) - will we need a pair of RAID arrays in order to really and truly trust these high capacity drives?
Linux with grub2 boots fine with a BIOS based PC with large drives using GPT partition tables. I have been doing it with hardware raid5 for a few years now. Might need to make sure /boot or / is all in the first TB or so to be sure the BIOS can read that part of the drive, but really who wouldn't on a drive that large anyhow.
Windows only requires EFI because Microsoft decided MBR partitions should always override GPT partitions while Linux and BSD went the other way (and hence work with hybrid setups on BIOS systems). Nothing prevents a BIOS based PC booting from GPT on a large drive other than Microsoft's bad design choice.
"Are there going to be many examples of people trying to boot from a 3TB drive?" Yes, of course - anyone who buys a PC from a mainstream outlet and doesn't understand the value of partitioning /usr away from the system will end up booting from 3Tb drives once they're commonplace.
We had this same problem when 16 bit Windows couldn't address the latest set of big drives and the world resolved it by migrating to 32bit Windows (there were patches you could apply if you insisted on staying with the 16 bit platforms IIRC).
In simple terms: the majority of home users and small, self-supporting businesses don't 'get' OS space vs. user data space - and why should they? We don't expect the average user to be aware of the inner workings of their cars engine and the same philosophy is applied to IT. We're in a good position to laugh about that but we really shouldn't: we (readers of El Reg generally make our living because we know more than the masses about technology - just as my plumber lives of me not wanting to know what happens when I flush the loo and my garage ripping me off ^H^H^H^H helping me when each service comes up.
"anyone who buys a PC from a mainstream outlet and doesn't understand the value of partitioning /usr away from the system will end up booting from 3Tb drives once they're commonplace"
By the time this becomes mainstream, it will be on a PC with 64bit Windows and the necessary BIOS to cope. A quick scan today of PC World, as a typical example, shows that even the top-end PCs have only 1TB drives - and they all have 64bit edition. It's at least 2 years before I'd expect 3TB drives to be in a PC World machine. It's at least another 2 after that before they get into the "cheerful" ones with a 32-bit build, by which time 64bit will be the mainstream product.
It will not be a problem.
Exactly, Annihilator - the OS will catch up exactly as it did in the 90's with the problems we had with RAM and so on.
Question: how would a home user backup 3Tb of video featuring Aunt Aggie and her mildly amusing dog? Mind you, how many backup what they have today? Best we avoid that whole issue, eh?
Paris: High time she got a look-in & anyway, can you /imagine/ what 3Tb of video of her would do to a grown man?
Ok so if I have this right, WinXP doesn't support these 3Tb hard drives, and 32 bit versions of Vista & Win7 only partly support it. If Microsoft can support it in the 64 bit versions then why not have support as standard in the 32 bit versions as well. Why does Microsoft have to make life more complex. (Mac OS & Linux can do it). I can't help thinking its things like this are unfortunately used by Microsoft as intentional planned leverage to encourage users to keep upgrading their OS (i.e. to force people to keep giving Microsoft more money).
This shows Microsoft are not just providing a service to their users. They are forcing their users, so is it any wonder users complain at these kinds of Microsoft tactics. Microsoft are paid a fee by us for their OS when we buy it, but then they fail to support it even when they could easily support it, because they instead want and expect us to buy the next version and so try to manipulate us into buying the next version. :(
I know at times its common to knock Microsoft, but there really are times like this when its justified by their actions. :(
the issue here is with the amount of space addressable by a 32-bit integer. A 32-bit integer is only able to directly access 2**32 blocks. Given a 512B block, that means that you can address up to 2TB of disk space. MacOS & Linux probably use the same type of hacks that are used to get 4GB+ of RAM into a 32-bit system. This would give them some additional space, but its only a patch really. The correct solution is to switch to a 64-bit+ system that is capable of handling such sizes natively. Given the track record on these types of things, its quite likely that these drive will work in a 32-bit WinXP system, but the OS will only see the first 2TB of space. The rest would just be ignored.
As to whether there's some reason why MS hasn't added the translation code into the 32-bit system or whether they're just "too lazy" to be bothered with it on an increasingly outdated architecture, I don't know.
Note: the 48-bit LBA scheme allows for drive sizes up to 128PB. A 64-bit scheme would support a drive with up to 8ZB of addressable storage.
Here's one way to get a 64 bit integer that works on a 32 bit machine.
__int64 myVar1;
myVar1 = very big number etc..
So you trying to tell me they cannot find *any way* to make a 3TB drive work on a 32 bit machine?!. I simply do not believe it.
It's nothing to do with than the CPU's ability to use big numbers, the limiting factor for disk sizes is the fact that the boot sector uses a 32-bit field for the count of the number of sectors on a disk, this limits the volume size to 2 TB regardless of whether you use FAT, NTFS of UFS.
So this means 64bit versions of Windoze can address exabytes of RAM but only terabytes of hard disk space!?!! And you still can't load the whole OS into RAM! What monkeys have they got writing their code, are they able of thinking even a few months beyond the planned ship date or their OS du jour?
These new disks are quite the unexpectedly jab in the eye for an ever more prehistoric software house who're lagging behind all their competitors...again!
Or is i3 3,000,000,000,000 bytes? I bet you it is not 3,298,534,883,328 bytes.
IT does not use SI units for data. There are 8 bits in a byte, then:-
1KB = 2^10 = 1,024
1MB = 2^20 = 1,048,576
1GB = 2^30 = 1,073,741,824
1TB = 2^40 = 1,099,511,627,776
If they are trying to "pull the wool over our eyes" again, they are short changing by 10% now. Someone really needs to take this through a court. In a lawyer 'blessed' land like the USA, I am surprised they haven't!
There are two kinds of K, M, etc., the binary kind, based on 2**10, and the decimal kind, based on 10**3.
Scientists talking about physical units use the decimal kind.
IT bods talking about RAM use the binary kind, but when talking about small units of time, they use the decimal kind (milli, micro, etc.).
Hard disks are measured in decimal units, which causes interest because it isn't until you get to GB sizes that you can actually have exactly n GB (decimal GB is 10**9, or 5**9 * 2**9, and you need a 2**9 to deal with the 2**9 sector size...).
Telecoms, however, introduces the third of the two kinds. An ADSL Mbps is neither 1,000,000 bps nor 1,048,576 bps. It is 1,024,000 bps, or a binary K of decimal K of bits per second...
hard disk manufacturers have ten bits to a byte, or something like that.
There are not two kinds of K, M, etc: there is the real kind, and the kind that are misrepresented by marketing departments.
Actually, it was HR departments that started it, quoting salaries in K for ,000s. They are just spinners and waffle merchants, like marketing people, too.
Yes indeedy, there are two kinds (plus the weird hybrid used by ADSL peeps). The "real" kind (i.e. the original sense, as used by scientists and anyone who talks about kilograms or kilometres) is based on 10**3. The "other" kind is the newer sense, based on 2**10. We can tell that it is this way round because all non-IT uses of these prefixes are decimal, while IT uses decimal for some things, binary for others, and both at once for the rest.
Of course, in reality, neither kind is real.
Where to start... first, it's 1kB, not 1KB. Second, 1kB doesn't equal 2^10, it equals 10^3 - because "kilo" is from the Greek word for "thousand", not "thousand and twenty-four". The kibibyte however, is 2^20 and is shortened to 1KiB. Extrapolate for the rest of your argument. Finally (and pedantically), a "byte" isn't necessarily 8 bits.
If they were calling this a 3TiB drive, you would have a point. But they're not, and you don't, so sit down.
MS started off with 640K of RAM being enough, and has continually ignored any idea of the future. They even ignore the current at times, such as PAE in the 32-bit desktop OSes being more or less useless, despite working fine for addressing more than 4GB RAM in their server OSes.
Here's a thought though - MS have a reason to cripple OSes... So they can sell new versions later.
http://physics.nist.gov/cuu/Units/binary.html
standards are created for a reason, lamons
if you dont know the difference between binary and decimal prefixes
thats a failed IT education right there, and you should ask your money back
(and beat up your dumb IT teacher too for being such a failure)
but i'm not surprised when there are still countries using body parts as measurement tools lmao.
De facto, AKA what everybody does and de jure, AKA what people on high horses think everybody should do. When the high-horsed people concoct ridiculous sounding standards, they shouldn't be surprised when everybody continues to use the old standard, especially when the only point of confusion is deliberately manufactured to sell storage.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
