Slashdot Mirror
WD's Monster 2TB Caviar Green Drive, Preview Test
MojoKid writes "Today Western Digital is announcing their WD20WEADS drive, otherwise known as
the WD Caviar Green 2.0TB. With 32MB of onboard cache and special power management algorithms that balance spindle speed and transfer rates, the WD Caviar Green 2TB not only breaks the 2 terabyte barrier but also offers an extremely low-power
profile in its standard 3.5" SATA footprint. Early testing shows it keeps pace with similar capacity drives from Seagate and Samsung."
Wasn't it only about a year ago that 1TB drives hit the market?
-jcr
The only title of honor that a tyrant can grant is "Enemy of the State."
Spindle-drives are inherently slow anyways, so I think the combination of a big, power-efficient drive (never mind the speed) for movies and an SSD drive for everything else is ideal.
What the hell do you do to back up your 2TB drive?
2 other 2TB drives?
Unless they're all the same model made in Thailand.
Shai Schticks:"You don't make peace with friends, you make peace with enemies"
Every time a new, larger drive comes out, people say, "That much data in one drive is dangerous!"
So here's what you do. Go buy ten 200GB drives. RAID them together. Who do you think will lose data, you, with ten times the possible failure points, or me with only one?
Just back it up, biznatch!
I never understood this argument. Say you have N drives each with capacity C/N (e.g. C=2TB, N=1 for this new drive, or C=500GB, N=4 as you prefer) and probability P of each drive failing in a given time interval. Your expected data loss is N*P*C/N, which is independent of N. So what's the gain from more drives?
Heck, assume you don't want the hassle of multiple partitions so you use logical volume management to concatenate the drives (simulating the larger disk). Since any failure kills the whole thing, it's even worse - N*P*C.
I guess maybe your are thinking of RAID5? But is this an enterprise-class hard drive? I'm not buying (or buying electricity for) 3x 1TB drives instead of 1x 2TB drive just to protect my PVR recordings. And since RAID (regardless of level) is not a backup, if the data is any more important than PVR recordings, you still need backups with or without RAID. So all RAID5 gives you is decreased time to recover from a broken drive, by making you buy a spare up front. Obviously decreased downtime is critical for an important server, but not for the vast majority of home PCs.
Except RAID isn't a backup, so your data isn't that "safe and happy".
If I have nothing to hide, don't search me
The problem with a larger drive is I fill it quickly. Should I buy a 2TB drive and use it to backup my already full two 1TB drives, or should I just add storage? Oh, the agony!
My RAID setup would use drives from different manufacturers and production lots, and contain hot spares.
Actually a multi terrabyte RAID 5 drive is a nice bit of the backup solution. No, it's not the be all and end all of backups. You still need separate completely off line and off site backups. But since modern RAID boxes can tell if a drive is bad, you get to look at the blinken light, go "oops", pull the drive and plug another one in. Wander off to Slashdot for a few hours and poof. Your data is back. No muss, no fuss. I like that part.
You can never be too rich, too thin or have too many backups. At least I get the chance to do one out of three....
Faster! Faster! Faster would be better!
Your argument would carry more weight if the manufacturers were doing this for the benefit of humans. In fact, they mix units - using the 1024-standard units for cache. Tell me mixing units is friendly! :)
W..w..W - Willy Waterloo washes Warren Wiggins who is washing Waldo Woo.
Let me explain this is detailed, complicated-for-you, terms.
Exponents are useful for counting possible combinations. Computers are logically binary and quantum. When dealing with storage, we use 1024, because it represents the number of possible configurations of a 10-bit sequence.
2^10 was chosen for convenience - it was close to 1000, which people are used to working with, and it provided a good separation between major units.
1000 was chosen by SI for reasons just as arbitrary, namely providing a good spacing. We have scalar units of 10 (decimeter, decameter, for example), but no one ever says "Go down the road 1.2 deca kilometers".
SI units (major units based on a factor 1000, with shitty units based on 10 for completeness) are for measuring.
Computery units, (major based on a factor of 1024, with minor units based on 2 as the basis), are for counting.
This is why clock speeds use 1000, not 1024. Clock speeds are traditionally measured, and not counted, and they do not operate on a binary quantum system.
This is why data storage is SUPPOSED to be described using 1024, while data transfer is described using 1000.
If you want to get down to it, all SI units are retarded, since the universe is quantum (it is). All measurements are merely inaccurate tools of convenience, and everything should be counted in universal quantums of space (Planck Length? I doubt it, unless it really is tortoises all the way down), time, etc.
To sum it up - SI is not right because it's "official". SI is WRONG for computer science. And if the universe is quantum, SI is technically wrong for everything. Calculus, too.
SI might be wrong for computer science, but SI prefixes have standard meanings. If we want prefixes that work better for computing (which we may well), then making new ones, just to be clear, is a good idea. Then if SI is wrong you don't have to use it, and you don't confuse everyone by using its terminology to mean something slightly different (which is much worse than using it to mean something very different).
Anyway, the power-of-two units make some calculations easier and many harder. Just because an N-bit MUX has 2^N inputs doesn't mean they'll all be connected to something. You have 4 384-byte memory modules, quick, how many kB? Um, what's 384/1024? 3/8 maybe? Having to mess with mutliplying/dividing by 1024 in the middle of back-of-the-napkin calculations where not every number is a simple power of 2 (even if many of them have lots of 0s at the end in binary, like 384 does) actually does suck unless you just give in and learn your multiplication tables in hex (if I was still doing driver programming I probably would have done just that).
Well binary units make sense for the size of solid state memories because they were almost always produced in power of two sizes (for good reasons, if they weren't the address decode logic would be a LOT more complex). Binary kilobytes also made sense for floppies since they were usually an integer number of power of two sized sectors.
And if your OS is already using a measuring system for memory and floppies doesn't it make sense to also use it for hard drives? MS clearly thought so. Unfortunately the hard drive vendors throught differently (whether for technical reasons or because it allowed them to advertise higher capacities is unclear) and so we ended up with two different systems in wide use.
Worse with each unit we go up the discrepancy gets worse. At kilo it was only 2.4% , at terra it's nearly 10%.
note: i'm known as plugwash most places but i screwd up registering that here somehow in the past and now can't register
Um, yeah, except that just about everything is stored on powers of two. This is as absurd as if hot dogs were sold in packages of 8 (or 1024) and buns were sold in packages of 10 (or 1000). AND they used the same term to describe both, until it got to the point where they could sell significantly less than was expected while using VERY small print to notify us of this change in wording.
There is absolutely NO reason to use base 10 numbering for computer memory of any kind, except that it allows manufacturers to use bigger numbers while selling less. The only mitigating factor is that now that they all do it, at least we're back to comparing apples to apples.
Sure I'm paranoid, but am I paranoid enough?
Now explain flash/solid state memory sizes and the "formatted capacity" of memory.
At a low level flash chips have a row/column nature similar to DRAM, but there is additional complexity because some operations target a larger "erase block" rather than an individual byte.
Some embedded platforms present a raw interface to the flash memory and require the host operating system to provide support for bad blocks and wear-levelling. These would be specified in "MiB" along with a certain allowable percentage of bad blocks.
However, the more familiar approach (e.g. in a SD card) is to include an embedded microcontroller that presents a logical block interface to the host. This controller skips over the bad blocks, and also needs to use some of the good blocks to keep track of the logical-to-physical block mapping. Here it makes more sense to use SI notation.
I just checked an "8 GB" microSD card and found that it presents a capacity of 7969177600 bytes to the OS (before partitioning). So 0.4% of the rated capacity is not available to me. This is consistent with the typical fine print on the package, e.g. "Some of the listed capacity is used for formatting and other functions, and thus is not available for data storage" (from Sandisk's website). If the manufacturer had sold it as an "8 GiB" card I would be more upset because that would represent a 7.3% capacity loss.
No one gives a shit whether your measurement system is 'correct', as long as it's consistent. SI is VERY consistent for a system that spans so many fields. That's why it's better than the US customary system.
There's nothing inherent to today's storage technologies that requires power-of-two capacities. We're not even using a fraction of the address space we already have, so sticking to a power-of-two size doesn't have any real benefits.
Oh, so Computer Science is so important that we get to invent our own units, and use the same names as established SI units? Please. If you want to use binary units because they are convenient, go ahead and do so. But DON'T call them "tera", "giga", or "mega"; these terms already have SPECIFIC meanings and you can't just hijack them.