Terabyte Hard Drive Put To the Test 376
EconolineCrush writes "As a technical milestone, Hitachi's Deskstar 7K1000 hard drive is undeniably impressive. The drive is the first to pack a trillion bytes into a standard 3.5" form factor, and while some may argue the merits of tebi versus tera, that's still an astounding accomplishment. Hitachi also outfitted the drive with 32MB of cache—double what you get with standard desktop drives—making this latest Deskstar a leader in both cache size and total capacity. That looks like a great formula for success on paper, but how does it pan out in the real world? The Tech Report has tested the 7K1000's performance, noise levels, and power consumption against 18 other drives to find out, with surprising results."
Data loss (Score:3, Interesting)
But on the other hand, a full-tower case loaded with those in a raid5 is enough to make me drool.
Re:Perpendicular (Score:4, Interesting)
it's been here for a while (Score:4, Interesting)
RAID 6 Please (Score:4, Interesting)
For example a 3ware 9650SE-8LPML can be had for as little $520.
Re:base 1024 (Score:3, Interesting)
Re:32 MB cache? (Score:3, Interesting)
Re:tebi? shut up. 1 terabyte drive still NOT here (Score:5, Interesting)
Let us take your absolutism to its logical conclusion.
Prima: I've got a huge car!
Secunda: Dude, I've got a huge cat!
* SUV-sized cat walks in.
Prima: Dude!
Secunda: (looking to camera) No, you see, "big" is an adjective, and must be read in the context of the noun it describes. A big cat is not the same size as a big car, or a big house, or a big boat. Prima: I see what you're saying. Similarly, a "kilo-gram" is prefixing the gram, a base-10 system, thus 10^3 grams; while a "kilo-byte", prefixing the byte, part of a base-2 system, refers to 2^10 bytes?
Secunda: Exactly! Humans, complex machines that they are, make use of context to bring out meaning.
Prima: But on Wikipedia it says this use is incorrect?
Secunda: Well, Wikipedia has the quality of a scientific journal... assuming submissions to scientific journals were all accepted for publication, and could be edited by anyone at any time.
Prima: So, the individual or group with the most amount of time ends up producing the predominant content?
Secunda: Exactly! The best way to confirm whether an article is likely to be useless is to read its talk page; in fact, you are more likely to learn from this page, as it illustrates the points of contention that one side or the other has tried to suppress.
Prima: So for the past two decades we have called 1024 bytes a "kilobyte", until one standards body associated with manufacturers of hard drives decided to redefine it...?
Secunda: Precisely. Worse, the previously unambiguous (outside of hard drive manufacturing) "kilobyte" is now defined as "1000 bytes". It'd be like renaming the mile to the "iMile", then stipulating that all future uses of "mile" should be based on the origin of the word - i.e. one thousand double paces.
Prima: But paces vary from person to person - it's like you're making an arbitrary change based in a tenuous argument that goes against the principle that language evolves other than by edict!
Secunda: Now you're getting the hang of it. Have you considered becoming a Wikipedia editor?
Tercera: Listen you two, either shut up or get a room.
Prima: Let's get some beer.
Secunda: Word.
* SUV-sized cat disappears in a puff of semantics, replaced with a slightly overweight puddytat.
Comment removed (Score:3, Interesting)
Re:tebi? shut up. 1 terabyte drive still NOT here (Score:2, Interesting)
The "mile" had been defined as 1000 double-paces since before the supposed birth of Christ. But then its meaning evolved in various contexts - the statute mile, the nautical mile, etc. Or, to use your language, "people ARBITRARILY redefined the mile". I hope that you maintain consistency with the original Roman definition when observing speed limits.
The "kilo", as you say, was defined according to the SI system in C18 to mean "1000 of". But then, as you barely well describe, its meaning evolved in a particular field. In fact, even better, it evolved within a specific context, so all its previous uses stand; and the redefintion was far from ARBITRARY, since powers of 2 make sense to use in binary, and powers of 10 usually don't.
Please try to get to grips with context in understanding language. It's a skill some engineers are very bad at; X in context A is not precisely X in context B. It never will be, because this sort of simplified reasoning abrogates the human brain's fantastic ability to recognise patterns without the need for identity.
Real-world use (Score:5, Interesting)
in front of thousands of people, where one mis-hap is already too much.
So far things have been flawless, and it has made a huge difference for me due to portability compared to anything else of the same capacity.
as previously this meant a two-drive combo with heftier power supply.
The weight and size make it easier to have it as a carry-on item, rather than in my checked luggage!
As far as performance, it has been able to handle 4 simultaneous 24-bit / 96 kHz audio tracks playing back with no hiccups whatsoever.
The drive-to-drive copying in Firewire 800 or SATA has been quite speedy and error-proof.... (copying 900 gig at a time is always a good test)
Dream come true if you ask me.... I still carry a backup anyway, LOL!
(ymmv(TM), batteries not included, kids don't try this at home, etc....)
Z.
Pretty small platters (Score:5, Interesting)
Seagate has announced (and released, I think?) their 1TB HDD with only 4 platters (cooler, quieter, less power, less weight, less cost to manufacture) that's 250gb a platter
Samsung have announced the F1 using 333GB per platter! 1.6TB if they copy Hitachi and slap 5 of them in a 3.5" unit - or rather 333gb single platter, light, cheap drives, be damned if anyone can find the F1 yet though
Solid State? (Score:3, Interesting)
Re:Data loss (Score:3, Interesting)
I go into bestbuy every once in a while just to screw with the geek-squad. One of my favorite things to do it read the specs of a system sitting on the shelf and ask someone if it would run that good with linux. Some would say anything if they thought you were going to buy it, some ask for the geek squad people to come over and field the question. And those boys tell you anything for any reason it seems. I have often thought about writing their answers down and publishing them somewhere. This reminds me of a time when a neighbor's cdrom quit working and he was told he needed a plug and play card (whatever that is) and it would cost $80 on top of the cdrom.
Re:O RLY? (Score:3, Interesting)
(The long series of calculations you have to go through in your post are the best argument for ditching the 1024*1024*1024 nonsense and just using thousands, millions and billions like the rest of the world.)
5.25"? (Score:3, Interesting)
Hard drives used to be physically much bigger [wikipedia.org], when the interface tech was "MFM: 5.25" diameter, and "Full Height" was about 3.5".
Physically smaller discs have faster access times and lower power consumption. But why not use larger discs for their higher data capacity, without wrapping each smaller chunk in the same electronics overhead for rotation and data transfer? And get the faster data transfer at the outer cylinders from their faster angular velocity?
At a guess, I'd say that a 5.25" full height HD could have 2.5x the 3.5" capacity per platter, and probably at least 5x the platters, for about 12x the capacity. The access times across the large areas would be larger, but for large files that wouldn't matter as much (as long as they're kept defragmented).
These truly "large" drives could be the best for archiving, thrown back in place after an emergency and gradually replaced with 3.5" disks (if necessary) as they continue to run.
We could have 12TB drives with the same encoding tech as these Hitachis. And they'd cost less per TB than the 3.5" ones, because they'd have more storage per overhead hardware. Where can I get one?
Re:Data loss (Score:3, Interesting)
It's also the way to go for speed - your controller doesn't have to calculate the parity bits for every write operation (yes I know the parity sum is simple - that doesn't stop it from adding a bottleneck).
The "bottleneck" of parity calculations is so small as to be irrelevant. Parity-based RAID levels are bottlenecked by the much higher number of physical disk operations, not the parity calculations.
Why didn't they compare it against 1TB Samsung ? (Score:3, Interesting)
Re:RAID 6 Please (Score:5, Interesting)
The Adaptec and LSI Megaraid cards are truly heinous. Just last week I had a system that wouldn't boot because the megaraid card decided that the NVRAM and on-disk settings didn't match... Even though the "force boot" option was set. Force-boot is supposed to write the on-disk config to nvram on a mismatch. As often as not, a machine with a megaraid card crashes on a single-disk failure instead of continuing to operate minus one disk. It'll reboot fine but not before you lose the unwritten data and deal with filesystem corruption. And God help you if a second disk develops a bad spot... It won't do the best it can to rebuild; it'll simply flunk leaving the good portions of the data unrecoverable.
I'll match Linux MD against those cards for reliability purposes any day. I wish there was some hardware I could buy that enhanced it with a battery-backed cache and parity acceleration. Then I could throw away the megaraid and adaptec cards.
The SmartArray cards are actually very good. Expensive as hell, but good. Sadly the primary configuration utility is on a CD instead of in the bios and some goober at HP decided to rig the disc so it won't boot on any hardware that's not HP/Compaq. Fortunately you can boot Knoppix, copy the linux config utilities and configure it that way.
Re:Data loss (Score:3, Interesting)
And then I met the beast known as the IBM Desktar 75GXP 40gb. I went through 5 of those, all the same drive (or rather, the same warranty) having to RMA it over and over. It was eventually replaced with a 60gb 60GXP. That one failed too. It's replacement is still working, but it makes a clicking noise every now and then that isn't the "normal" read noise - I don't put any data on the drive. Just apps that I have backup install media for. There was eventually a class action suit brought against IBM for these drives. They were just terrible.
I'll still never buy anything that says Desktar again, and despite the possiblity of the crash effecting it, my time spent as a computer tech while in college has prompted me to never buy Western Digital again either. Though I'm sure there are other reliable brands, I've developed a very, very good respect for Seagate drives. They never tend to be the fastest, but they've never given me any trouble.
Re:tebi? shut up. 1 terabyte drive still NOT here (Score:3, Interesting)
Why fight the rest of the world over this? Now that we have binary prefixes, let's use them! This idea that metric prefixes are base 10 in networking and base 2 in storage is embarrassingly inconsistent. Let binary prefixes mean binary, and let metric prefixes mean base 10! Just because we did it one way in the past doesn't mean it is the best way to do it now! This is engineering, not religion.
Re:Data loss (Score:3, Interesting)
Re:kanashhk shhk shhk (Score:3, Interesting)
Hitachi_Hard-Drive_Project_-_Noriko_Version.mp3 [odeo.com]
Written by James Postlethwaite, whose home page I can't find, and made entirely out of hard drive failure noises (Hitachi provide a nice set of wavs [hitachigst.com]).