Slashdot Mirror

The Register's testing method (copy 3GB from/to drive) is questionable at best. The drive has been reviewed elsewhere more comprehensively, with a not so one-dimensional result.

Right now I'm using a laptop with 2 hard drives. Last summer, 2-3 months after buying it, I upgraded one of the drives to a 160GB Seagate Momentus 7200.2. At the time it was the fastest drive on the market. It's since been surpassed by the 200GB Hitachi Travelstar 7K200. With a transfer rate min 40-71.5MB/s max http://www.storagereview.com/HTS722020K9A00.sr?page=0%2C1 it's as fast or faster than these fancy new "oh so awesome" SSD drives. The only benefit would be power savings. Flash sucks. The Adobe driven ads and the RAM. It's too slow to be considered in the average device. Even the latest model 32 and 64GB flavors only come in at about 55MB/s http://www.tomshardware.com/2007/12/17/solid_state_drives/page5.html, but it is a steady transfer rate media wide. My C drive currently has 40.8GB used on it. That could be trimmed (and 12GB of it is system restore garbage), but I'd still probably want a 32GB SSD drive to run Vista (XP doesn't like my hardware at all, too lazy to force it). I might even be able to get by with a 16GB drive if I uninstalled some large programs. But still let's go with your 8GB OS drive solution. The cheapest SATA SSD drive on NewEgg is an 8GB 2.5" .1 watt $240 drive. Or I can get the fastest magnetic drive on the market, 200GB 2.5" 1.5watt idle/5watt load 7200rpm Hitachi 7K200 for $165. Seeing as my laptop has been plugged up 99% of its life and 2.5" drives are virtually inaudible, I know which I'd choose. Which would you choose? The cheapest 16 and 32GB SATA SSD drives that I'd HAVE to have to run my laptop are $350 and $700 respectively. For either of those single drive prices I can buy one of the fastest AND one of the largest magnetic drives out right now, the $165 Hitachi 7K200 AND any of the 3 $140-$170 320GB drives (the largest out is not 500GB). Cost is the reason we can't/shouldn't have both. Flash drives are still niche tech. And that's why you can add one if you have to have it. But why would any retailer push anything they'd have to sell so hard? Dell: "Here, this $500 option will give you 30 minute more battery life per charge." Customer: "But isn't it less space? Why would I want that? Why does that cost more?" Dell: "Yes. Because it's good for the environment? It's fancy technology!" Customer: "I think I'll just get the $150 extended battery option." Straight off Dell's website while customizing a laptop. "Reliability: 64GB Solid State Drive and Speed:200GB SATA Hard Drive (7200RP [add $1,000 or $30/month1]" The only reason I can think they wouldn't offer a smaller drive is it's really just too small. And it probably would cut into their profits.

Go read Storage Review. They have performance and reliability databases.

Last I looked, Seagate were doing pretty well reliability wise, and their latest ES.2/7200.11's were doing amazingly well with NCQ/multi-user IO and STR. Hitachi (IBM) do pretty well for single user performance, and er, a quick glance at a pricelist shows they're in no way overpriced. ISTR a spate of WD failures being reported in forums which put a few people off them, along with a few 5-percentile entries in the reliability db (i.e. less reliabile than 95% of the drives in the db), but take that with a grain of salt.

Personally I just buy Seagates. Never had a problem with their internal drives. The external "OneTouch" 750G Seagate I've got is passable, but I don't really like USB or Firewire (does USB Mass Storage really have power management control? The drive spins down every 10 minutes or so and it'd be nice to stop it without hacking something to write to it when it's there, or opening it so I can use ataidle(8) on it).

Assuming the [Green Power] also shares such a seek time, that leaves us with 15 ms [measured access time] minus 9.5 ms [assumed seek time] which equals 5.5 ms, almost exactly the rotational latency associated with a 5400 RPM spindle speed.

[I]t's easy to convert [WD's values for average rotational latency] to revolutions per minute, or RPM. 5.6 milliseconds of rotational latency works out to about 5,400 RPM, which just happens to be the low end of the GreenPower's spindle speed range. Western Digital says that's by design; the latency spec it lists in the GreenPower's data sheets is merely an estimate based on the spindle speed range of the drive.

No, HDD's are hermetically sealed. They have to be.

Actually, no. Look at any consumer drive and you'll see a hole with a sticker besides it with an arrow that says "do no cover this hole". This is to equalize pressure, otherwise the cover would bulge and warp. (There is a filter under the hole so particulate matter can't get in). Yes, there are some very special disks used in the military and NASA and the like that are hermetically sealed mainly to keep moisture from condensing on the platters, but you don't have one of those drives, trust me.

Spacing between head and platter can be in the microns. Turbulence from the rotation of the disk actually aids in making sure the head does not contact the disk.

Yes, and the turbulence (it's actually the boundary layer) that keeps the head off the disk is air, which gets less dense with increasing altitude, and so it gets thinner, ergo crash. What I meant was that perhaps the designers knew this and included a shut-out circuit that detects low pressure and doesn't allow the drive to spin up.

See: http://www.storagereview.com/guide2000/ref/hdd/op/heads/opHeight.html or Scott Moulton's fantastic drive recovery videos: http://www.myharddrivedied.com/presentations.html for more information

One might be humidity, another the decreased frictional constant of the air around you, viscosity of the air is what the internal fans rely on to cool the circuitry and heat producing chips, that the air is thinner means decreased cooling capacity.

Actually humidity drops with altitude. I agree that there is much lower cooling capacity at higher altitude, although it has nothing to do with viscosity or a frictional constant. Air is less dense at high altitudes, so the volumetric heat capacity of air is lower at high altitudes, so for a given volume of air that your fan moves, a lower mass of air is blown over the chip at higher altitudes, and there is less cooling. So sure, it could be a heating thing, no doubt. Whether it be for hard drive protection or thermal reasons, the cut-out would probably be tied to a pressure sensor, since at high altitudes, the thing won't even turn on (it doesn't freeze up like an overclocked computer). Many camcorders, digital cameras, etc. won't either. I have a few solid state gizmos that don't like high altitude either. In any case, the SSD can go higher.

... (and yes, when I said I was a biologist who works in high altitudes I should have said I am a biologist who works on heat balance physiology of high altitude organisms)

Performance wise, once you switch to desktops, you are able to use performance drives like Western Digital Raptor WD1500.

http://www.storagereview.com/php/benchmark/bench_sort.php

Compare the Western Digital Raptor WD1500 No NCQ to the Western Digital Scorpio WD2500BEVS with NCQ (250 GB SATA). The Scorpio consumes a lot less power, but isn't that much quieter. The Raptor has about 2.5x the performance.

SSD wins on noise and power, and the Raptor wins on price. Depending on the application, either could win in performance.

Seagate apparantly did a good job with the firmware on their new 7200.11's with NCQ; there's at least one benchmark on Storage Review forums, showing serial transfer performance of 9 seperate reads running upwards of 80MB/s, while every other tested drive struggles to get past 20.

Might be a useful excuse to upgrade to a 1TB drive if it helps more common use-cases.

Interesting that you work for Seagate, but don't know that the ES.2 achieves over 100 MB/s on the outer edge, and over 50 MB/s on the inner. (According to StorageReview, anyway.)

"Attempting to recover data packed at a density of 1 GB/sq.in. from a disk spinning at 10,000 revolutions per minute where the actual data is stored in a micron thin layer of rust on the surface of the disk is manifestly impossible."

It's not always iron oxide . Sometimes it's Cobalt

According to Storage review they are made of aluminium (eventually alloys) or glass (eventually mixed with ceramics.)

I like the tech report's personality better, but not really surprising results, IMHO. Old news from May:

http://www.storagereview.com/HDS721010KLA330.sr?pa ge=0%2C7

It seems possible that this magnetic affectation could be a cause of spontaneous damage the hard drive servo information.

This would cause one of the clicking-type malfunctions which you described, as that "clicking" you hear is the noise the head assembly makes when the drive is rapidly moving it back and forth across the platter attempting to get a fix.

You couldn't get 16GB 3.5" HDDs in 1996. My 1999 machine came with an 8.4GB drive and that was pretty standard.

I've been trying to understand why caching to a USB flash drive - instead of a hard drive - is a performance win. USB's max transfer rate is 60 MB/s; in practice it's more like 30-40 MB/s. If I have a moderately-non-sucky HD, it has a minimum transfer rate of 30 MB/s. A good SATA drive can do 60-90 MB/s: (according to StorageReview).

I suppose the USB drive does eliminate seek time and rotational latency, but it's not obvious to me that this is a performance win. Has anyone seen benchmarks?

And yes, TurboMemory is on PCIe, so it's different. But the parent brought up Vista's caching to USB...

What are you on? Spindle speed is far more important for sequential transfer rates than areal density. It's also the biggest factor for seek time, which can impact a server with lots of concurrent I/O significantly more than raw transfer rate.

500GB/7200RPM SATA drives have a minimum sequential transfer rate of around 35-45 megabytes per second, with the maximum possible somewhere around 75 megabytes per second. 73-147GB 10,000RPM SCSI drives have comparable minimum transfer rate performance, but with maximum rates of 80 to nearly 100 megabytes per second. For new 15,000 RPM SCSI drives, the *minimum* sequential transfer rate is 70-80 megabytes per second, topping off at between just under 100 and 135 megabytes per second at the edge of the platters. You can't buy 15,000 RPM SATA or IDE drives - that kind of performance is only available in SCSI, SAS, and FC drives.

What are you on? Spindle speed is far more important for sequential transfer rates than areal density. It's also the biggest factor for seek time, which can impact a server with lots of concurrent I/O significantly more than raw transfer rate.

500GB/7200RPM SATA drives have a minimum sequential transfer rate of around 35-45 megabytes per second, with the maximum possible somewhere around 75 megabytes per second. 73-147GB 10,000RPM SCSI drives have comparable minimum transfer rate performance, but with maximum rates of 80 to nearly 100 megabytes per second. For new 15,000 RPM SCSI drives, the *minimum* sequential transfer rate is 70-80 megabytes per second, topping off at between just under 100 and 135 megabytes per second at the edge of the platters. You can't buy 15,000 RPM SATA or IDE drives - that kind of performance is only available in SCSI, SAS, and FC drives.

What are you on? Spindle speed is far more important for sequential transfer rates than areal density. It's also the biggest factor for seek time, which can impact a server with lots of concurrent I/O significantly more than raw transfer rate.

500GB/7200RPM SATA drives have a minimum sequential transfer rate of around 35-45 megabytes per second, with the maximum possible somewhere around 75 megabytes per second. 73-147GB 10,000RPM SCSI drives have comparable minimum transfer rate performance, but with maximum rates of 80 to nearly 100 megabytes per second. For new 15,000 RPM SCSI drives, the *minimum* sequential transfer rate is 70-80 megabytes per second, topping off at between just under 100 and 135 megabytes per second at the edge of the platters. You can't buy 15,000 RPM SATA or IDE drives - that kind of performance is only available in SCSI, SAS, and FC drives.

What's interesting to me is that neither of these papers mentions the issue of pre-installation handling. The good folks over at Storage Review seem to be of the opinion that the shocks and bumps that happen to a drive between the factory and the final installation are the most significant factor in drive reliability (much more than brand, for example).

The google paper talks a bit about certain drive "vintages" being problemmatic, but I wonder if they buy drives in large lots, and perhaps some lots might have been handled roughly during shipping. If they could trace back each hard drive to the original order, perhaps they could look to see if there's a correlation between failure and shipping lot.

-R

What's interesting to me is that neither of these papers mentions the issue of pre-installation handling. The good folks over at Storage Review seem to be of the opinion that the shocks and bumps that happen to a drive between the factory and the final installation are the most significant factor in drive reliability (much more than brand, for example).

The google paper talks a bit about certain drive "vintages" being problemmatic, but I wonder if they buy drives in large lots, and perhaps some lots might have been handled roughly during shipping. If they could trace back each hard drive to the original order, perhaps they could look to see if there's a correlation between failure and shipping lot.

-R

Reliability data for desktop drives can be found here: StorageReview

Unfortunately you need to register to see it, but that's what bugmenot was invented for ;)

I'm afraid I disagree with you though - I think Google and everyone else should make their data public. It would save everyone a lot of pain, and make the manufacturers of unreliable drives actually improve their game.

Places like tomshardware only review one drive, and most likely for a couple of days - so I don't think that's really their responsibility.

When a friend broke down, she asked the breakdown man who came what were the most reliable cars. He said he wasn't allowed to comment but that "he carried no honda parts". I guess the same thing applies here - Google won't say, they'd get sued.

On the other hand, hard drives change so much that this year's model will be totally different design and mechanics than next years, so blaming (say) IBM for its crappy deskstar range should not be reason to blame their (ok, Hitachi's) current line.

If you do want to know more about which drives are best - check out storeagereview and enter details of your drives to their reliability database.

Another site that might be useful for the average consumer is http://www.storagereview.com/ :)

Forgot to say the first 15K drive was 2000... an article with the dates and speeds.

The html tag parsing mangled my reply when I used a less than symbol. Fixed reply:

What was that Hi-Cap driver that I installed to allow access to the 300GB drives I have installed in the boxes then?

Something that bypasses the 128GB limitation of single partition size by doing a little trickery. I trust you noticed that you have to partition the drives into less than 128GB chunks.

There are no "inexpensive" ATA PCI cards that work for a mac. They are starting out at $65+ everywhere I have seen them. Cards for PCs don't work.

Uh, $65 is inexpensive. More inexpensive than the only alternative you implied ("buying newer stuff [from Apple]").

You're wrong that I'm wrong. I have installed large drives on countless boxes. They may require drivemagic or a BIOS update, but I have yet to see a PC that was limited by the hardware.

If they require additional software/drivers, that's the same trickery as Hi-Cap.

In any event, the fact that the ATA controller on early G4's didn't have 48-bit LBA/Large Disk support isn't a "bug". Earlier ATA controllers didn't have such support. (And if you think Apple purposefully did it when disk sizes were commonly less than 40GB with designs on "forcing" people to upgrade when >128GB disks became available, you're deluded.)

What's really amusing is you seem to have no problem doing essentially the exact same solution you're using on the G4 on PCs.

More info:

http://www.48bitlba.com/
http://www.seagate.com/support/kb/disc/tp/137gb.pd f
http://www.storagereview.com/guide2000/ref/hdd/bio s/sizeGB128.html

What was that Hi-Cap driver that I installed to allow access to the 300GB drives I have installed in the boxes then?

Something that bypasses the 128GB limitation of single partition size by doing a little trickery. I trust you noticed that you have to partition the drives into There are no "inexpensive" ATA PCI cards that work for a mac. They are starting out at $65+ everywhere I have seen them. Cards for PCs don't work.

Uh, $65 is inexpensive. More inexpensive than the only alternative you implied ("buying newer stuff [from Apple]").

You're wrong that I'm wrong. I have installed large drives on countless boxes. They may require drivemagic or a BIOS update, but I have yet to see a PC that was limited by the hardware.

If they require additional software/drivers, that's the same trickery as Hi-Cap.

In any event, the fact that the ATA controller on early G4's didn't have 48-bit LBA/Large Disk support isn't a "bug". Earlier ATA controllers didn't have such support. (And if you think Apple purposefully did it when disk sizes were commonly less than 40GB with designs on "forcing" people to upgrade when >128GB disks became available, you're deluded.)

What's really amusing is you seem to have no problem doing essentially the exact same solution you're using on the G4 on PCs.

More info:

http://www.48bitlba.com/
http://www.seagate.com/support/kb/disc/tp/137gb.pd f
http://www.storagereview.com/guide2000/ref/hdd/bio s/sizeGB128.html

HDs are comparatively slow and flash drives are approaching the big enough state where they could replace them in e.g. laptops and workstations.

Typical overall file transfer speeds are about 3 MBytes/s. The highest current overall file transfer speeds are about 10-25 MByte/s. Older, "full speed" 12 Mbit/s devices are limited to a maximum of about 1 MByte/s.

Your hard drive, assuming a recent model of at least 7200 rpm, is capable of 50-135 MB/s (see Maximum Transfer Rate. Even connected to a PCI IDE controller, it would still run at up to 133 MB/s. So it's definitely not your hard drive responsible for limiting it to 12-25 MB/s, assuming you aren't doing anything else disk intensive at the same time.

Storage Review has had a Hard Drive Reliability Database for quite some time now. I've listed all my drives there (Not only the ones that fail). I recommend you check it out :

http://www.storagereview.com/map/lm.cgi/survey_log in

It would be nice to know whether the levels were A-weighted or linear

SCSI drives have been aimed specifically at the server market for years. Their electronics have been specifically geared towared performance in that area.
SATA is aimed for desktop systems. Almost no early SATA systems had NCQ, although more and more newer models have it now. It's been tacked on, not central to the design for decades like in SCSI.

Just take the baddest SATA drive around, the Western Digital Raptor 150. A 10,000 RPM drive with NCQ (Note, the earlier 36 GB and 74 GB Raptor SATA drives did not have NCQ. It's a new tack-on)

1) Look at it's benchmarks vs a bunch of SCSI drives for single user benchmarks.

Wow, look. It not only blows away other SATA drives, but kills those expensive SCSI drives. This is why the uninformed grandparent things SCSI is 'ancient' tech.

2) Now look at it's benchmarks vs a bunch of SCSI drives for multi user benchmarks.

Notice in the top graph where it slaughters every other SATA drive out there except for it's earlier sibling, the Raptor74. It's obviously pretty much the best SATA has to offer.

Now look at the next graph where it is compared to those 'Ancient' SCSI drives. Those SCSI drives beat the hell out the piss-poor excuse of NCQ that that high end SATA drive has.

That's why SCSI is still king for server work.

I've got a Raptor 150 in my home gaming machine. It's great. But I'd never think about it in my RAIDed servers at work. It just couldn't come close to the job the SCSI drives do.

SCSI drives have been aimed specifically at the server market for years. Their electronics have been specifically geared towared performance in that area.
SATA is aimed for desktop systems. Almost no early SATA systems had NCQ, although more and more newer models have it now. It's been tacked on, not central to the design for decades like in SCSI.

Just take the baddest SATA drive around, the Western Digital Raptor 150. A 10,000 RPM drive with NCQ (Note, the earlier 36 GB and 74 GB Raptor SATA drives did not have NCQ. It's a new tack-on)

1) Look at it's benchmarks vs a bunch of SCSI drives for single user benchmarks.

Wow, look. It not only blows away other SATA drives, but kills those expensive SCSI drives. This is why the uninformed grandparent things SCSI is 'ancient' tech.

2) Now look at it's benchmarks vs a bunch of SCSI drives for multi user benchmarks.

Notice in the top graph where it slaughters every other SATA drive out there except for it's earlier sibling, the Raptor74. It's obviously pretty much the best SATA has to offer.

Now look at the next graph where it is compared to those 'Ancient' SCSI drives. Those SCSI drives beat the hell out the piss-poor excuse of NCQ that that high end SATA drive has.

That's why SCSI is still king for server work.

I've got a Raptor 150 in my home gaming machine. It's great. But I'd never think about it in my RAIDed servers at work. It just couldn't come close to the job the SCSI drives do.

StorageReview agrees: http://www.storagereview.com/articles/200601/WD150 0ADFD_7.html

I didn't know Fireballs could do network storage!

Fireballs have been able to do network storage for at least a decade...

People often forget there is a considerable difference in the reliability of ATA drives versus SCSI. If you are going to use some sort of ATA based SAN be prepared for disk failures much sooner than if they were SCSI.

This is not necessarily true. It all depends on how your network storage is being used. SCSI drives are built and firmware'd for the sole purpose of running a server, and they consistently beat any ATA drive (be it IDE or Serial) when it comes to server performance and reliability. ATA drives just aren't built to handle the sort of usage a server requires--note that this isn't a reflection of quality, but of purpose. But a file server (which is the only thing the SAN would be used for) requires much less robust firmware than a server housing MySQL, PHP, maybe a CRM suite, e-mail server, etc.--and so ATA drives shouldn't immediately be ruled as less reliable. The maturity of the technology plays a more important role than the interface.

This drive has also been reviewed by SilentPCReview for those of us, who are more interested in noise than in performance. Another 500 GB Caviar model, the WD5000YS, was covered by StorageReview -- IIRC, the differences between those two drives are in the firmware.

The SATA specification is able to reach 300 MB/s, but single harddrives can't even saturate UDMA. Gigabit NICs only reach 125 MB/s, so one could have almost two of them in a 1xPCIe slot. Good point with the RAID cards, but isn't that what 4xPCIe is for? :)

As for PhysX, it doesn't really need a lot of bandwidth. All it does is send updated positions for all the objects, which can be done with one 3x4 matrix per object (4 * 12 bytes). Even on a scene with 5000 objects running at 100 updates/s it would only need roughly 24 MB/s (48 bytes per object * 5000 objects * 100 hz = ~24 MB/s), well within the limits of PCI. PCIe would probably help in the initial setup of the scene, but the PPU will get bottlenecked by the PhysX processor before it is slowed down by PCI.

Having big 16x slots with less than 16 lanes does sound like a useful feature, I had no idea macs had something like that.

The major advantage of SCSI or FC over SATA is its performance under heavy multi-user load. I refer you to test results at http://www.storagereview.com/articles/200601/WD150 0ADFD_6.html

True, the IOMeter performance of the drive revewed, and most SATA drives under deep queues isn't as good as the more expensive SCSI/FC drives out there but looking at this fact in isolation gives a skewed picture.

Most servers operate with a queue depth of 1 most of the time and that's especially true in a small office. If you are looking to make an array that performs well under queue depths of 128, you are generally dealing with many thousands of users and therefor probably not designing for a small, inexpensive system. If you are looking to see how fast a bunch of people can transfer medium/large files to/from a machine, don't look at I/O's per second in deep queues, it's not applicable. The High-End DriveMark is really more appropriate there.

Even so, even if you somehow have an environment where you are going to be pushing queue depths way up, my argument still holds. In an array with 8 of those Raptors reviwed with a queue depth of 64, each drive can read about 190 I/Os per second for a total of 1500 I/Os per second. If you have 4 of the very best 15,000 RPM Maxtor drive you'll be pushing about 400 I/Os per second per drive with a depth of 128 for a total of 1600 I/Os per second. That's appears to be about the same speed but even that comaprison isn't fair. Those are both 150GB drives and the access pattern is random across the whole disk. With 300 GB of mirrored data (the most that 4 150 GB FC disks could hold) you'd only half fill the 8 SATA disks. Random seeks across half the disks surface are much faster than random seeks across the whole disk. The 8 SATA disks would far outperform the 4 FC disks in I/Os per second, and the speed increase would be even more dramatic when dealing with low queue depths and sequential transfers.

IOMeter is not a typical light-medium load/file server usage benchmark and it's results should not be interpreted as such. For example, for a file server running Samba, the application is single threaded so the queue depth for it will never exceed 1. IOMeter performance is really best used when an entire set of disks was dedicated as storage for a database with a large number of concurrent users. That's what these SCSI/FC drives are designed for and it also why you don't see FC in small businesses and education. Where you see it in use is when there are thousands of users hitting a system and you will lose millions of dollars in productivity if the array fails. That's when you buy FC and that's why the prices are so insane. These high-end drives don't do anywhere near as well when used in single threaded/file server rolls and are often even slower (look at the High-End DriveMark results in the article you linked.)

The performance of drives under deep queues has little to do with the interface and much more to do with the speed of the internal queueing mechanisms and the drives seek time. There's nothing magic about SCSI or Fibre Channel that will make a drive faster. It's simply high end heads, seek arms, and powerful controler logic designed for heavy concurrent use.

For a light dubty, high speed, milti-purpose array I'd suggest a nice 8 drive SATA array with a good 8 drive RAID controller. It will be bigger, faster, cheaper and able to handle more load than if you spent the same money on FC equipment. If that speed and size isn't required, go with a 4 drive SATA array. You would probably be amazed at how fast something like that is if you get a good controller and good drives.

"Storage systems designed with Fibre Channel have almost no advantages over SATA based ones and cost much more. How is that sensible?"

The major advantage of SCSI or FC over SATA is its performance under heavy multi-user load.
I refer you to test results at http://www.storagereview.com/articles/200601/WD150 0ADFD_6.html/

The OP

There's a thorough comparison of SATA vs SCSI drives at http://www.storagereview.com/articles/200601/WD150 0ADFD_6.html. For multiuser use 10000 RPM SCSI (or its close relative, Fibre Channel), typically performs 150 - 200% more I/Os per second than the WD Raptor 10000 RPM SATA drive. SCSI drives also come in 15000 RPM flavors, which widens the performance gap even more.

One way to curb some of the cost, I might add, would be to switch to something like RAID 5... you won't have as high throughput, but you'll still see performance gains and end up with more usable drive space. The throughput likely won't be your problem, anyways... typically it would be the drive's ability to handle multiple simultaneous requests, which heavily relies on low access times (which is why SCSI dominates in this type of environment).

Here's a quick reference of some IOMeter benchmarks using a file server test pattern. You'll see what I mean. Wealth of info on drives on that site.

Most won't notice any speed difference when moving from PATA to SATA. On PATA you typically have two harddisk one the same controller, but that hurts performance when using both disks at the same time. With SATA this is not a problem, assuming you have enough SATA connections available. NCQ may reduce desktop performance, and is most usefull for server like environments. For more info, search Storage Review

Quote of the StorageReview article;

Cosmetically speaking, the Raptor X pays homage to the DIY crowd that enjoys things such as clear PC cases with... well, a transparent cover. Though the actual top plate remains a jet-black aluminum compound, the portion that actually sits atop the spindle assembly, constructed from polycarbonate, permits a clear view of the drive's interior. To further the product's unique appearance, WD investigated incorporating an LED into the device but ruled it out due to engineering constraints.

Long story short, no LED for yuo

(and about the price thing, you're paying the window+jet black casing $50 extra over the regular version of the drive)

As most people know, movie props are often made of common items and then painted, dressed-up, etc, but you don't often notice them as such. Now here's how this is related to the subject at hand (don't mod me off-topic just yet).

I'm not sure how many non-geeks (or even semi-geeks for that matter) know what the inside of a hard drive looks like or what the parts look like. But there, inside Darth Vader's helmet... the one used as a prop in ROTS... are two stacks of hard drive head arms. They just look like some high-tech gizmo to give it a cool futuristic cyber look.

I wonder how many people actually saw them and recognized them for what they are. I have no idea if they can actually be seen in the movie or not. I just though it was kind of cool that there are hard drive parts inside Darth Vader's helmet.

-S

The new Raptor - it's far faster than even several SCSI drives (in real world, "gamer" stuff), it's got more than 2X the storage than its predecessor, and it's coming at a price point of $300-350. (that's just $100 MSRP higher than the 74GB version).

New Egg has the drive for $295.

This performance comparison has the drive's gaming performance... It's as fast or faster than 15K SCSI drives! (single user, single app performance on this page, BUT - the article does have full benchmarks)

And that's just ONE drive. So, RAID 0 is probably pretty rockin.

And if you're already a Raptor user, it's my bet that this will lower the price of the other models. It's time to get my RAID 0 on!

A warranty is a good measure of how reliable a manufacturer EXPECTS a drive to be, not how reliable it actually is. The deathstars, for example, were much more failure prone than IBM expected. There is no way to know about issues like that from warranty information. MTBF numbers usually given out are the same thing, not based in actual data but based on engineering estimates.

To know how reliable a drive is, you have to know actual failure rates. Only the manufacturer is typically in a position to accurately measure those and they pretty much never give it out without an NDA or court order. We on the outside are left manually piecing together the data using methods like The storage review drive reliablity survey:

http://www.storagereview.com/map/lm.cgi/survey_log in

which attempts to gather accurate statistics from large samplings from users. This seems like a lot of work but hopefully it will pry the window open and convince manufacturers that it won't be the end of the world if people know how reliable their drives actually are.

I'm a consultant, and let me tell you that hard drive failures know no brand loyalties with modern drives. We see more dead laptop drives than desktop drives, but that's because of the rougher treatment.

But even the (relatively) large numbers of drives we see is anecdotal. Let's hear from the *real* experts:

http://faq.storagereview.com/tiki-index.php?page=B randMostReliable

-R

I guess this is a good time to bring up the storagereview reliability database. It's the only third party tracking of HDD reliability that I am aware of. Whenever I buy a new HDD or have one die or taken out of service I go to storagereview and update my profile. Other people may not be so reliable, and people with problems are probably more likely to report then happy customers, but it WILL give you a good idea model vs model of the reliability of a drive.