Slashdot Mirror
Hitachi's 500GB SATA-II Reviewed
Doggie Fizzle writes "The specifications for the Hitachi Desktstar 7K500 are impressive. 500 GB of disk space, 16 MB of cache memory, and 3.0 Gbps of transfer speeds are about as good as you are going to get in today's hard drives. The only category that might be rivaled is transfer speed, but that would require RAID or an Ultra320 SCSI drive to do so. This BigBruin review matches it up with some Seagate drives to show off its performance."
With 7K500, I believe that the RPM which is not the best compared to the 10k drives avaible today.
What does your Credit Report look like?
3 gbps? Is that 375 MB/s? IDE/SATA doesn't support that! What's the point?
For %95 of the population, do the specs of the latest and greatest matter?
Yes, yes, I know we are the 5%.
-m
How can you trust a 500GB drive from this dubious "Hitach" newcomer who is obviously just typosquatting Hitachi's reputation?
They dont like you calling it that. There's not SATA 2 standard as yet.
It's instead, SATA 3Gb/s. Most motherboard manufacturers jumped the gun however and invented their version.
Matt
Anyone know what the reliability on this line of drives is? After my experiences with the IBM line, I'm hesitant to buy anything with Deskstar on it. Just recently I replaced my ATA Deathstar (AGAIN!) and Hitachi sent me what looks to be a rebranded IBM. Same model, could be the same drive for all I know.
I'm guessing the newer Hitachi line of SATA drives no longer carries the IBM Deathstar plague, but I'd like some assurance before plunking down cash on it. In the meantime, I'll tolerate the performance losses of a Seagate if it means there's a better chance of keeping my data.
My friend Ben had one of the infamous Deathstars; he had to pay shipping to IBM after it died, and the replacement died within one month, and the next replacement within two months, and the next replacement within two months, and he had to pay shipping and go without a hard disk each time. I think his fourth or fifth Deathstar finally lasted him a decent little while, or he got another disk.
Anyway, if IBM thinks that's acceptable, I won't ever be buying one of its disks.
How long would it take the average slashdotter to fill that puppy with pr0n?
when my Deskstar drive crashes after only a week of use.
Hypocrisy is the 8th deadly sin.
I still can't get past the stigma of these drives.
Its like hearing about a new form management tool from Claria.
liqbase
Alright, so I'll lose 500GB of crap when the deathstar craps out
Live life to the fullest. It's not that life is short, but that you are dead for so long.
Gravitation is a theory, not a fact.
SATA-II indeed supports that. So does the disk. From cache.. No way it reaches more than 50MiB/sec from the platters, which is what counts. So I think it should be dead easy to rival speed with raid. My 6 year old IBM 18.2GB UltraStar drives read 25MiB/sec, so 3 of them would outperform in read/write. But would not take that much data...
So, indeed, it is a large disk, but it is not extraordinarily fast. Of course, bigger disks means more data per second, since the platter size is the same. Then data has to be packed more densily, and more data passes under head per second. So the disk can read more, in a sequential read.
Assembling etherkillers for fun an profit
Does this thing gets perpendicular? :-p
Hmm More pr0n - first think popus in my mind :D
The important thing is not to stop questioning --Albert Einstein.
I have the 160GB deskstar.
Little did I know when I bought it that every 15 minutes it would make a loud screeching noise as it performs a self-check.
There's no way to turn this off and it's über annoying. It's a lovely drive in all other respects, but I won't buy another unless I know for a *fact* it doesn't behave in this way.
--
Toby
Isn't that one of the more important thing with HD's nowdays? Sure, speed is nice but it wouldn't matter much if the HD crashed after two years. Having a HD that is only three years old, but "already" started to report SMART warnings. It makes me wonder how reliable the HD's are of today. I heard alot of people having HD's crash on them, and most of the time it's HD's from the last three years. Have they become more unreliable? (And yes, i'm going to replace the HD on this computer soon. I start to notice a few oddities with it.) At least this HD have three years warrenty, which is nice. Then, my HD started to act funky just when the warrenty went out...
If they turn out to have a premature failure bug, they'll become known as the Hitachi DeathStar. (The last disk I had with some designed-in irony like that was the Quantum Fireball.)
(obligatory)
more room for pr0n!!!!
Lame recently registered site that goes directly to a add hosting service.
LATENCTY LATENCY LATENCY LATENCY LATENCTY LATENCY LATENCY.
LATENCY is what is causing the slow performance of hard drives, who cares what the MB/s is (its good enough) its the latency that kills you more than anything RAID will not increase LATENCY. RAID can only make things more complex and make it worse (no system can be 100% efficient). RAID can increase MB/s but as I've allready said that isn't a big deal. What we need is lower latency Hard drives. We have enough storage. I don't need 500GB I want good latency.
At one point, the IBM/Hitachi ATA drives had command tag queuing that allowed for performance that was more in-line with SCSI. The link is /.ed - does this line of drives bring the command queuing back? I've been looking for some new drives for servers, and these sound mighty nice, even if they are "deathstar's".
Jerry
http://www.cyvin.org/
I have lots of various IBM/Hitachi drives in my systems including 160GB, 120GB, 80GB, 45GB, and 40GB, as well as some 9GB Seagate drives. None of the Hitachi drives make any significant noise at all. They are incredibly quiet. The 9GB Seagates on the other hand hum very loundly, but none of my drives make any sort of screeching noise.
Morphing Software
Try smartctl.
/dev/hda should do that (yes, even in Windows).
smartctl --offlineauto=off
having one of those brain things is a leading cause of headaches and hangovers and frustrated libido. no thanks!
If these drives allow companies to do business more efficiently, consumers may realize a benefit if these cost savings enable competition to lower prices.
Is it me, or have advancements in harddrives been slowing down? 400GB has been king for a over a year, and only two manufacturers seem to even have a 400GB offering. Just a few years ago, it seemed that everytime I turned around bigger drives were coming out. Have we finally hit some kind of limit for magnetic storage?
That means it will take at least 22 minutes to backup all my pr0^H^H^Hfiles. I don't have that kind of time!
3 gbps? Is that 375 MB/s? IDE/SATA doesn't support that! What's the point?
SATA-II indeed supports that. So does the disk. From cache.. No way it reaches more than 50MiB/sec from the platters, which is what counts.
So true. I'm not really understanding the point of having such a large on-drive cache. I think the money is better spent on adding RAM to the main computer because the OS does a lot of caching too. A multi-tasking OS on hardware that has DMA capabilities seems to make large on-drive caching unnecessary. It seems like the testing I've seen of real-world programs on the drives with different size caches confirm this train of thought.
Oftentimes adding huge caches to CPUs nets almost no speed difference as well, and CPUs are far faster than hard drives.
suggested that this drive get very hot indeed, as it is 4 platters not 3. Didnt really seem worthwhile to me, as heat is a major cause of HD failure.
Same here, even though i know it's for the best if it does that.
For the record, i've recently purchased a 80gb 7200rpm Hitachi Deskstar after a faulty psu burned my old trusty Seagate. I needed the drive in a hurry, and was a little bit uneasy with the Hitachi drives (you know, ex-IBM...), but after 6 months of non-stop server use i have to say they're excellent. Fast, reliable, and very quiet - not as much as the Seagate Barracudas, which you couldn't tell if they were running or not, but close.
Seagate is still my #1 drive brand, but from my own experiences and what i've heard from them, Hitachi is a close second choice.
YES! Commies Love NATALIE!
Now, imagine a world where your company name is spelled corrctly on Slashdot's main page!!!
Aggressively working towards deconstruction of Apples' fan base (if it happens one more time)
Why the hell use a format that does not run properly on the PC. It's a slap in the face
http://hi-techreviews.com/reviews/7K500/hitachi.ht ml
You get (more than) 3 7k250 250GB drives for the price of one of those 7k500 drives, so they are not very attractive for building a very large archive.
Oftentimes adding huge caches to CPUs nets almost no speed difference as well, and CPUs are far faster than hard drives.
I must disagree strongly here. An old, but still valid example is the difference in performance between the AMD K6-2 and K6-3 processors. The former has 16 KB of L1 cache; the latter has 16 KB L1 and 256 KB L2. All cache runs at the processor frequency.
In general floating-point number-crunching applications, I have observed a 2x to 2.5x improvement in performance with the K6-3 at the same processor speed. Sure, results may depend on your application, but don't dismiss processor cache as useless. There's a reason you pay more for it.
It is pretty annoying how, to this day, so many people get so excited about max theoretical bus speeds and confuse it for actual performance. The only time you will get anywhere near 3Gbps is during a transfer from the drive's cache. Otherwise, you are limited by the media rate and head seek time of the drive. These are the primary factors in real world performance. The bus speed is rarely a bottleneck for hard drives except in situations like SCSI where you would be putting many drives on a single bus. I bet this new Hitachi drive would perform nearly identically on a SATA 1.
-matthew
"THERE IS NO JUSTICE, THERE IS ONLY ME." -Death
Yes, we have hit a limit. The magnetic bits are too small now to continue without a change in technology.
h ead/pr/PerpendicularAnimation.html
100 gigabytes per 3.5" platter is about the max we seem to be able to do.
The answer? Get perpendicular!!!
http://www.hitachigst.com/hdd/research/recording_
Modesty is one of life's greatest attributes
If I recall correctly, that's the firmware fix for the old "will head-crash on inactivity"-bug that hit the rest of us on the older models :-\
Don't worry, you'll be having a forced upgrade soon unless you're lucky.
:-/
Or fancy spending £70 on the DHL shipping to Netherlands
HITACH SUNFLOWER now this and 8 gigabyte ram disk cards, a passive backplane, a 3ware sata raid controller, tyan thunder, no more latency, lots of cache orders of magnitude faster than disk i/o, not only that, but now you need to build another one and cluster them. geez lois what are they going to think up next?
True
You misunderstand me. I did NOT dismiss the processor cache as useless. Just that after a certain point, there is no point in increasing it.
That is an old and invalid example, the invalidity isn't necessarily because it is old.
K6-II / K6-III doesn't apply because the basic cache arcitecture is different. for II, the cache was expected to be on the main board, outside the FSB, for III, the cache was put on-die, inside the FSB.
For more relevant comparisons, see the comparison between 512MB and 1MB cache Athlon64s, or the Pentium M with 2MB cache against the same clock Celeraon M with 1MB cache.
[brag]My quad WD 10k RPM Raptor array pwns this thing.[/brag]
-Very steady- around 80MB/s sustained throughput from the beginning of the array to the end. Peaks of 104MB/s. Troughs of like 72MB/s.
According to this review this 500GB Hitachi starts out at 65MB/s and trails off to a pathetic 35MB/s.
Question everything
OK, I mispelled Celeron.
Hitachi makes Deskstars. Hitach, as one can clearly read above, makes Desktstars.
You can hold down the "B" button for continuous firing.
After a certain point, yes. That point in processor cache is about 2MB. That point for a disk depends on usage, but we're not in danger of hitting it for the vast majority of uses.
http://www.electronicsweekly.com/ (UK) reported about this disk half a year ago! http://www.electronicsweekly.com/articles/article. asp?liArticleID=38427
Can't believe you didn't get any funny mods. Hitach Desktstars :)
https://www.gnu.org/philosophy/free-sw.html
If the disk did not have the cache, it would have to write directly to system memory. Only one device (CPU, hard disk, video card, etc.) can write to main memory at a time, so this would significantly increase the amount of time you have to wait for memory acccesses from the cpu - everything would be waiting on the disk to stop filling cache.
Now, as to why such a large cache - the lack of power back-up makes it less useful. It would be better if there was less memory on the disk, and a lot more on a controller.
Correct me if I'm wrong, but doesn't SATA-I support 150 MB/s, and SATA-II doubles that, which means that it would support 300 MB/s, not 375?
I have personally had several IBM/Hitachi drives fail, loosing GB's of data. Even all these years after the desk star fiasco, their drives still suck. My PowerBook HD died in May. Yup, it was a Hitachi.
I don't think I'd use that drive if you gave it to me. That's a deskstar, aka "deathstar" in the sysadmin circles. I have a STACK of those drives at work, all doing the same thing. Power them on, and you hear a chirp-click-chirp-click that just repeats. The drive never spins up. Tried replacing the controller card on them, that's not the problem, it's something inside. That stack is actually not all of them either - a class action suit was just recently settled and we submitted claims for another stack of deathstars.
We might have one deathstar in the building that still works, and if I find it I'm replacing it. Save yourself the headache, do not buy deathstars. When maxtor bought quantum, maxtor adopted quantum's designs, and now produces decent drives. Hitachi bought IBM's drive line, but they just inherited the crappy deathstar design and that's what they're selling.
The only model of drive I have seen perform as bad as a deathstar is the old Quantum Fireball 6.4gb's, which tended to smoke their spindle motor controller IC. At least those you could swap controller cards and save your data.
I work for the Department of Redundancy Department.
I don't want the fastest or biggest disk.
Just the quietest.
I think the money is better spent on adding RAM to the main computer because the OS does a lot of caching too.
I believe that while the system's cache excels at saving disk reads (in fact it's faster and more effective than the disk controller's cache ever could be) the disk controller's cache can offer significant acceleration for disk writes. The system's cache can only postpone writes, not accelerate them. With a controller cache, data may be dumped to the disk controller at the full bus speed, rather than being limited by the speed of the spinning metal.
I think you'll find that when comparing drives with identical specs apart from the controller cache (Western Digital, for example, has offered "Standard" 2MB cache and "Special Edition" 8MB cache versions of otherwise identical units for some years), the drive with the larger cache does indeed get better benchmark scores.
And 8MB or even 16MB of RAM for the disk controller is very cheap these days. Skimping on that cache wouldn't save enough money to make a significant upgrade to the system RAM.
Recent Hitachi return policy prevents me from even considering this line of HDs.
I attempted to return a failed IBM Deskstar last year only to be told I would have to return it to the US, not the Canadian centre I had used in the past.
I explained repeatedly that I had always returned HDs of all makes to Canadian centres and that it was prohibitively expensive to ship a DEAD HD to the US.
Hitachi didn't care. I have never bought a Hitachi drive since and never will.
I have been using Seagate HDs because of their 5 year warranty and have not had a single failure to date. Seagate = cool, quiet and reliable.
Goodbye IBM/Hitachi, Hello Seagate.
Brian
I don't have a Brian.
Who is Brian?
Latency generally refers to a delay. Look it up... harder. Here we're talking about how long it takes for the drive to respond to a request for data. It's relative to how fast the disk spins, how far the head is from the requested data, etc.
These big drives seem destined to be available only with fast IO interfaces. Makes sense when the data object is consumed at high-bandwidth, like HDVD video. Or when many concurrent streams are accessing the data, like a large scale (many users) Video On Demand app. The large storage capacity is reflected in the large transmission capacity: scaling up current data apps to more users or better resolution data.
But the biggest change we have right now is the ability of individuals to have lots of items of the same old size. People watching their own videos from their own libaries of hundreds of movies. Listening to their own songs from their own libraries of hundreds of thousands of songs. Those apps require huge storage, like hundreds or thousands of GB, for a single person. But they therefore don't require high bandwidth transmission. A 5400RPM EIDE drive is plenty fast enough, but it still needs 500GB capacity (which density might require the higher RPM, but not the faster interface, caches, etc). And for consumers, the overhead for IO bandwidth is a waste of money. As is more than maybe 2 or 3 drives for RAID failover, which also demands cheaper drives.
Hitachi's 0.5TB SATA-II drive is targeted at datacenters and multiuser servers, with money for bandwidth. So where are the cheap, huge, Personal Computer drives? Say, 500GB EIDE for $250?
--
make install -not war
They claim up to 10x density... Is it theoretical or real? I don't know, but its what they claim.
So, instead of 100 gB / platter, thats 1,000 gB/ platter. With 4 platters (they use 4 in 3.5" 400gb maxtor drives), that would be...
4 Terrabytes in a single 3.5" Drive. (mb/mib/Mb/mB blah blah blah conversion loss)...
Tada.
Modesty is one of life's greatest attributes
Dunno, maybe turning off the caps lock helps reducing latency...
It's not a loud screeching sound, more like a little chirp. You are too sensitive, Miss Princess and the Pea. I have the same drives (two of them actually) and I don't even hear it over the cooling fans.
The "chirp" is thermal recalibration. All hard drives do it. You can't turn it off and you don't want to.
for my...friend's pr0n collection. Everything would fit within 3 of these drives.
From the spec sheet:
Sustained data rate (MB/sec) 64.8 - 31 (zone 0 - 29)
Actually it isn't 375MB/s. AnandTech just had a few articles on this: http://www.anandtech.com/storage/showdoc.aspx?i=24 50
You loose 20% of the bandwidth for parity, so it ends up being 300MB/s.
HJ
That's what that sound is! I have a two, 250GB Hitachi drives in RAID0 and every once and a while I'll hear them do a "SCREEEEeeechcreeech" sound. I wasn't sure if they were failing or if that was a normal sound.
Yes, it is slightly annoying at worst. Most of the time it's not very loud at all. However, I have a co-worker who has an old Mac G3 with a hard drive SO DARN loud you can hear it continuously make a high pitch sound. I hope the thing dies soon so I don't have to hear it anymore.
Latency is all about an initial delay in reading data.
It's the same with RAM, networks, Drives.. you ask for data, and there is a slight delay while the system gets itself set to give that data to you. Usually, once you've started retrieving that data, the rest comes really quickly as its cached, or otherwise stored sequentially.
ie. Imagine a drive with a file stored bit-by-bit in sequence. You ask for the file, once the heads have moved to the right point, the drive will read all the bits and return them to you. Latency is that initial delay.
Now, imagine you're asking for 10 files, each a tenth of the size of the original.. you won't be able to retrieve them all in the same amount of time. So a drive with higher latency will take correspondingly longer to get you those files than one with lower latency.
This is also why CAS latency is important in RAM, and why gamers will spend loads extra on CL2 modules. Also why getting 1 large amount of network data is very mich faster than getting it in smaller chunks.
64MB being on the rim of the disk. 50MB would be a good average I think, but yes, I see your point.
3Gbps is indeed 375MB/sec... But yes, sata is 150MB/sec, so I'd guess that 300MB/sec would be more correct.
Assembling etherkillers for fun an profit
many accesses are random, and access time is determined by the time it takes to move the heads to the new track, plus the time it takes for the desired sector to rotate under the heads. RPM makes a difference in the latter, since it is on average 1/2 the rotation time, regardless of the number of bits on a track.
"National Security is the chief cause of national insecurity." - Celine's First Law
Disagree. We had two defective 46.1Gb 60GXPs, both failed in service - bye bye 100Gb RAID0 of renders :(
If that wasn't bad enough the way IBM behaved about it was. We had to pay to return them, and then they shipped us refurbs as warranty replacements. Which duly failed as a RAID1 a few months later.
When you add that to a whole rack load of distinctly flaky IBM 18.2Gb 10k SCSI drives I had at work means I wouldn't touch any of their hard disk products again if you paid me!
Hmm, I've had fine expirence with some of their newer products (touched them again this year). In particular I've tried the 80GB Deskstars, which have performed fine in a lot of machines I've built. And the scsi UltraStar drives has also worked fine for me...
Assembling etherkillers for fun an profit
He is right!
NCQ actually increases latency in desktop usage profiles due added overhead. In only can show its benefits if there are long queues of outstanding transactions, which rarely happens outside server usage.
HI O WISE PRINCE. WHT TOOK U SO DAM LONG?
so you triple the number of drives and you can read stuff faster...
shocking... now lets triple the number of these drives. oops still twice as fast.
than you simply dont know how silent good hardrives are (hint: non of those you mentioned are quite by any means).
One Hitachi is louder than my 4 samsung Raid together.
HI O WISE PRINCE. WHT TOOK U SO DAM LONG?
Lose. Not loose.
I'm not normally a grammar nazi. However, this mistake really irks me.
What's the point of using RAID if you are going to put samsung drives on it? Your data is already as good as gone.
I was a fan of the Deskstar (Deathstar) line until my 75GXP started making strange noises. Pulled it before an actual failure (no data loss).
My IBM Thinkpad came with a TravelStar, slow, but it worked.
Then Hitachi Came out with the 7k60, the 7200 RPM 2.5" drive! It was bliss. A few weeks ago, disk errors all over the place.
Now I've been waiting a month for Seagate to ship it's new 7200 RPM 2.5" drive.
No more Hitachi for me. The drives aren't designed well. That's the bottom line. Fast? Yes, definately. But not reliable.
Hello to Seagate, home of the 5 year warranty.
500GB:
On the one hand it allows me to store more pr0n.
On the other hand, it allows me to lose more pr0n when the HD crashes.
Decisions, decisions.
Well, there's spam egg sausage and spam, that's not got much spam in it.
Is there any legal use for such space ?
My point was indeed that if read performance was the important thing, it would be cheaper to use multiple, small drives. Which holds true. 2x200GB would cost less, nearly same capacity and quicker (normal sata drives, 7.2KRPM). So this ain't a competetive disk for price/capacity ratio.
Assembling etherkillers for fun an profit
Too bad you can't delete your own post. You go out of your way to ridicule someone while making yourself sound like an idiot. Latency is, in computer terms, the time it takes between a request and a response. The hard drive is the slowest piece of equipment on a computer. Therefore, it has the highest latency. The latency IS the bottleneck you describe. When you want to read a file, the latency of the hard drive is the time it takes to locate the file and send it to the hard drive controller. Mirroring the hard drive, as you mention, is a way to decrease latency for read times. RAIDs (especially RAID 5) are ways to decrease latency and increase data throughput. Faster processors won't help you if you can't feed data to and from them fast enough to use them to capacity. More RAM is helpful and more processor cache is as well, but the biggest bottleneck on a computer is the hard drive. In general, the larger the hard drive, the faster because all things being equal, the outer track of the disk will read more data per spin on a larger disk (higher density) than a smaller one. If you really want to increase speed, get multiple processors to work on multiple tasks simultaniously and get a RAID 5 setup to pull and store data faster. Of course, get at least a Gig of RAM to work with, but remember anything over the size of your OS and the files and programs you're working with that loads into memory is just wasted RAM. RAM is great to free you from a SWAP file, but beyond a certain amount, you're getting diminishing returns. I have an Athlon 3400+ with 1 Gig of RAM and 2 SATA 200 Gig drives. When encoding video for DVDs, I use data on one to write to the other and my processor doesn't even hit 80% capacity, nor does my RAM max out. My bottleneck is the hard drives themselves, but it's actually faster for me to have the OS and the raw data on one disk with the output file on a seperate disk than to have a 2-drive RAID mirror in this situation. I'm switching to RAID 5 with a promise raid controller soon, I think. Much lower latency. Remember... Lower latency = faster data transfer.
Don't you remember the ibm-disk that broke after some moths? IBM sold his hdd-branch to hitachi. So Hitachi is NO NEWCOMER. This are she very same poeple that created the unreliable ibm drives. So even if they bring out a 500 exabyte hdd with 500 gb of CACHE for only 200 (hey, I have this "currency sign"-key on my keyboard, so i use it. ;)
i would not buy it!
I rather wait some years before looking for them again...
Any sufficiently advanced intelligence is indistinguishable from stupidity.
The cost/benefit point for processor caches is about 1MB or less. A 2MB cache CPU is maybe a percent or two faster than the same CPU with 1MB CPU. A 1MB cache CPU is maybe a percent or so faster than the same CPU with 512k cache. Except for commercial uses, the benefit of doubling cache is almost certainly not worth the extra money that the CPU makers ask.
For drives, from what I've seen, doubling cache rarely nets more than a low single digit in real-world speed increase.
Huh?
I looked for this command on a Windows XP machine, an OpenBSD machine and a Linux (Gentoo) machine. Command not found on each one.
For those too lazy to do their own search, the smartmontools package can be found here. Even works with OpenBSD!
Your dum.
If you plan on running Adaptec SATA controllers and Hitachi SATA drives, do some extensive testing first. We ran into a confirmed nasty bug between the two. One that Adaptec and Hitachi have no intention of fixing.
Oh yeah, and there is no such standard as SATA-II. Serial ATA International Organization. SATA-IO. And its not supposed to be binary for TWO as a previous poster has surmised.
File not found.
But I figured out your link:
Linky!!!
Or you can type in http://smartmontools.sourceforge.net/
Karnal
Compression free video? I rip or download files for all the TV DVDs i've got, and more. Average 1hr show is around 350MB in a DivX or Xvid .avi. Seems like a lot of space, but its really only about a fifth of the space of whats on the DVD. Compression is good these days, but its still not as crisp as off the discs themselves. Sure having the DVDs is nice, but you've still gotta pick one, take it off the shelf, put it in the player, switch your AV stuff to it. Whole lot easier to treat the DVDs as a backup to sit on a shelf for display, or even in a box, and then just being able to call up anything in seconds off the HD. Its nice to hit a button and have my playlist of a couple dozen series come up on shuffle. Its like adult swim, sci-fi channel and TNT daytime all rolled into one whenever i want. I have 2 250GB SATA drives in RAID0, and infact ordered another 100GB drive the other day to move my music onto (65 or so gigs of legally ripped CDs, 300+ CDs @256kb MP3). If i had the money, would i like 2 of these? Hell yes, double my storage without taking up any more room in my already full case. I will admit that all my stuff isn't 100% legal, but thats more out of practicality, simpsons DVDs are only up to season 5... and fair use vs DMCA with the issue of me being too lazy to rip stuff myself...
"Sic Semper Tyrannosaurus Rex."
Ah yes, RAID0. Ideal when a single Deskstar isn't unreliable enough.
echo -e 'global _start\n _start:\n mov eax, 2\n int 80h\n jmp _start' > a.asm; nasm a.asm -f elf; ld a.o -o a;
As nice as that is, it's still a Deskstar.
How cool are you going to feel when your 500 GB drive dies?
Actually, you can turn it off. I have a 160gb sata deskstar flashed with different firmware. No meow-sounds att all :)
/ushac
It's not official though, and I can't seem to find a link to it anymore. Try googling a bit. I think someone said you can mail their customer support and ask for it too...
Are we at a length (eg, 2M) where externally connected SATA devices are a realistic option?
1M is probably do-able, but once you factor in interface port to end device and the actual cable runs involved (port to external connector, external connector to external device, external device port to actual device connector), 2M seems like a minimum for no-fuss connections with breathing room.
Look at it this way. The enginearing cost to get cache onto the drive at all is probebly the same for 16MB as for 2MB.
The other cost to consider are the RAM chips themselvs and in the topsy turvy world of Tiwanease suplyers and volume priceing, a 16MB chip may be cheaper than all it's smaller siblings.
--= Isn't it surprising how badly I spell ?
Where does tha 20% that you set "loose" go?
Just like I'd never buy one of those unreliable Japanese cars. Why, my brother owned a Datsun in the 1980s and he had nothing but problems with it. I'd never buy an unreliable Japanese car after that bad experience.
please write it sereval times, plus in CAPITAL LETTER, so i can read them better and understand you... ohh.. nevermind
But I think my sig may be relevant here...
I have a 30 GB hitachi that I had been running full time (about a year) in my ancient tangerine iBook, using it as my home web server. That drive always had a tendency to be noisy when I bought it, but I had bought it used, and getting it into the iBook the first time was such a pain I didn't take it back.
Every now and then it would buzz and click, like a refrigerator starting its compressor, but lower volume, and repeating.
About a month back, it wouldn't quit the buzz-click, so I figured it was time to shut down and hope I could recover my data over the weekend. Wouldn't boot up, so I started calculating how much data I'd lost. (Also prayed a bit, because there was some that is not replaceable.) When I had time on the weekend, I opened the iBook and swapped the old 6 GB drive back in. (Also Hitachi, but it usually is not noisy.
I hooked the 30GB drive to the ATA-to-USB converter that I had plugged the old drive into (one of those shirt-pocket-case kits), and, indeed, it did not spin up. Tried freezing it and also tapping it, no effect.
So, since I had nothing to lose, I broke the seals and popped the cover. The platters are a nice silvery-aluminumy color. Carefully avoiding touching the platters themselves, I rotated them manually by the hubs. The first spin, I could feel the flat bearing or stuck grease breaking, then they spun easy. So I plugged the ATA-to-USB interface back on, plugged the USB cable to the iBook, and it powered up and mounted just fine.
My son enjoyed watching the platters spin before I replaced the covers, but he was disappointed that he didn't get the magnets out of there after all.
Got all my data out, and they still spin. I may be able to use it for moving non-critical data, I suppose. Definitely not for anything critical.
bad sectors, disks that won't spin up, more bad sectors. And the interface bottleneck.
Where I have to use ATA technology, I (now) make sure I back up regularly. Back up that notebook every night. Actual cost requires adding a spare drive for every drive in use.
For anything even halfway important, give me SCSI, even if it means buying the controller separate.
That stink isn't going away just because they did a company change.
Now all I need is $1600.00 to outfit my box with enough hard drives to hold my DVD collection. At least all the drives will be internal instead of some kind of Firewire boxes.
I drank what? -- Socrates
Every which way but loose.
I just had a Toshiba laptop drive die... just after its 1 year warranty was up... but its prob for the best, as it is replaced with a 5year warrantied Seagate (got some other Seagates, too, for the servers...) ... but now I'm wondering... Ought I to do some EXTRA backing up almost 5 years from now?
The Admin and the Engineer
Secret.net.au have an excellent script to work out $$$ vs GB (based on their current price list):d dcostpergb.pl
http://www.secret.com.au/cgi-bin/computers/util-h
You have a sick, twisted mind. Please subscribe me to your newsletter.
Nobody is going to want this drive because Hitachi and IBM before them have a bad reputation for reliability. This 500GB superdrive (no, I dont respect apple's lame trademark on this name) doesn't do anything but get coolest geek on the block status, so 5% is really too high.
Segate and WD are what's hot. Hitachi is more expensive on average so they don't sell too well. If someone wants it cheap they'll go with a MDT, Maxtor, or a "refurbished" (cough) WD/Segate.
What do I care? Market factors. Having only 2 legit HD companies isn't very good for consumers, but fortunatly prices are still low for the present. If Hitachi were reading this forum (HA!) I'd be happy for them to work off my criticism.
"And we have seen and do testify that the Father sent the Son to be the Savior of the World"
1 John 4:14
When you do 5.1 production, espically for DVD-A, it gets real large real fast. The final product is about 100MB/minute. The tracks to produce it can easily go over 1GB/minute.
I have a folder on my disk where I'm just playing around, not even doing any serious production, with a couple of 5.1 mixes in different formats. It's 6GB.
I'm sure HD video production is even worse, but I don't do that.
500 marketing gigabytes = ~465 real gigabytes.
Prevent email address forgery. Publish SPF records for y
Use gentoo only if you -know- what you want. It's like saying "I build my own car but it does not have cruise control..."
emerge sys-apps/smartmontools does the trick...
Why would anyone post a link to a site that isn't ready yet? Thanks for the popups and popunders, though.
Hey, I chuckled -- give the guy a break. :)
with a Masonary nail through it. Over the years I've had problems with most brands (or helped out people who have), so am a little bit dubious of the quality of all drives.
I'm working, drives are cheap, I just wish they'd stop crashing. Anybody think there's room in the market for a drive that costs 50% more, but gives you double your money back if it fails in warranty? (I know that would be open to abuse, but something where the manufacturer has an incentive other than the postage cost of a repaired drive).
I'd just finished reading this when I saw the parent post ;-p& mode=classic
http://ars.userfriendly.org/cartoons/?id=20000801
I too had one of the infamous Deathstars. When it died after a mere 4 months of service, I went ahead and bought another drive, but still returned the first one. Since it was within the first year, I was able to return it to the point of purchase, and they shipped it back with their weekly "batch" of dead drives (as any large distibutor will have). I got a replacement two weeks later, and it has seen far more than the 4 months the first one did, with nary a hiccup. It's sitting on a shelf because it's relatively small now, but last I checked, it worked just fine. I'll be using it soon to restore a Norton Ghost image that has some critical data, since I'd rather now overwrite any of the partitions on my main drive.
Not every HD return and replace story is a hideous one.
Mal-2
How is the Riemann zeta function like Trump rallies? Both have an endless number of trivial zeros.
...we only had punch cards for storage.
And we liked it!
With an unrecoverable read rate of 1 in 10^14, it would only take 200 full disk reads to encounter an unrecoverable read, statistically speaking.
As disks continue to get larger, it seems more important that read reliability increase as well.
Tiller's Rule: Never use a word in written form that you've only heard and never read. You will end up looking foolish.
Actually I use Gentoo because I -know- what I -don't- want. Cruft. Simple as that.
It's like saying "I build my own car but it doesn't have a 285 hp V8 because I don't feel like wasting money on pouring gas into it."
Also, Gentoo is the easiest system to work with. It is the best 'noob' distribution, as long as you are willing to get dirty. Simplicity is very attractive.
What's a 'MiB/sec'? Is that a millibyte per second?
antipaucity
RAM is cheap but HDD manufacturers ask for a relatively hefty premium for the privilege of this trivial upgrade: a 64Mbits (8MB) RAM chip costs around $2 while a 256Mbits (32MB) chip costs under $4... yet HDD manufacturers ask $10-$20 more for the 16MB version. For that kind of price difference, they could offer 32MB and still make some extra profit.
Then again, having huge on-board caches would increase the risk of losing data when power randomly goes out, which would seem like a valid reason to hold back on caches... let an auxiliary RAID/storage controller do extra caching for battery-backed-up external boxes.
SATA-II != 3Gbps.
SATA-II is only the name of the new standards comittee for SATA devices. Devices are not guaranteed to support NCQ or 3Gbps unless the device manufacturer specifically says so.
As for why 3Gbps != 375MB/s, this is because 3Gbps is the wire speed but this wire uses 8B/10B code expansion to facilitate clock recovery and error detection, which is like having 10bits per byte. Add out-of-band signals, CRCs and the rest, the usable speed might drop to 250MB/s... full-duplex though.
My 80GB WD drives peak around 60MB/s, my 120GB drive peaks around 65MB/s, all three being nearly two years old.
50MB/s is history, 70MB/s peak (outer cylinders) is probably common with current (2005) 7200RPM/3.5" drives... but I have not benched my two newest drives yet to verify that.
I buy a lot of big drives (of all brands).. Whatever the biggest on the market is when I need to add space (several times a year). I'm finding that drive reliability has gotten really shitty with most of these drives lasting less than a year. I'd like them to stop working on making drives hold more and concentrate on making drives more reliable. A drive should last at least three years under heavy but reasonable usage.
At what price learning? At what cost wisdom? The price is a man's peace of mind, and the cost is his life.
"MiB/sec" is a Mebibyte/sec. Mebibyte = 2^20 bytes. Megabyte = 1,000,000 bytes. But you knew that, lol.
Simplicity is very attractive.
That's why I love to fuck retarded chicks.
The cost/benefit point for processor caches is about 1MB or less. A 2MB cache CPU is maybe a percent or two faster than the same CPU with 1MB CPU
Given the size of information being handled, 1 or 2 Mb is too small for cache to really make much difference. If you open a document or even a web page you can easily use more than 2 Mb RAM. The exe code can stay in cache - very helpful - but data is still slower. Cache >= 64 Mb would be really fast though clock speed will suffer.
A compromise worth considering - more than one point of memory access so that multiple CPUs on one motherboard can work together quickly. I tried some high speed multitasking on a Pentium D and found that it was bottlenecked by the inability to load data from the memory as the separate cores try to work. High speed processors are mostly used to handle large volumes of information rather than arriving at one answer from a small amount of data.
Know your pads. One time pad: good for cryptography. Two timing pad: where to take your mistress.
Interesting comments. We have been using the 400GB Deskstars for about 1 year and they work just fine. We have 50 of them that we use with our backup server (www.idealstor.com) and we rotate the drives on a daily/weekly basis like we used to with tape. I am still waiting to get my hands on the 500GB's to test them out. We may be looking at getting some of the Seagate drives with the 5 year warranty because i have heard good things about them as well.
Hitachi makes their own stuff. Heck, they dont even have the same cache algorithm as IBM anymore. (Given my job, I would know)
+1 funny, -2 overrated. Life isn't fair.
Yeah.. right right - sorry.
one of the things I am noticing is that people are having a general whing about how crappy the deskstar is. They are even having a go at IBM who hasn't made the drive for a couple of years now.
I am going to presume that most of you don't have access to production servers or are not SA because if you were you would know that drives from all manufactures fail regularly! When ever you buy a SAN they will give you a mean time between failures for your drives. Same thing for servers etc... Remember that is only a "mean" time which means that out of the all the disks of that model the manufacture sells over 50% will last that long, the rest could last longer or shorter.
Personally I have owned IBM (deskstar), seagate, maxtor and western digital drivers are all of them have failed at one point in time or another (the deskstar has been the most reliable).
My point is that if you rely on a HD (especially non raid), no matter what manufacture, as the only place for your important data then you deserve to lose it. At home I backup to external HD + other server. At work I have a DR(co-lo) site + tape + raid(I raid everything).
It said "windows 98 or better" so I installed Linux