Flurry of Hard Drive Reviews

Sivar writes "After a long hiatus while setting up their new testbed, StorageReview.com has released a number of reviews of the latest hard drives, including Hitachi's Deskstar 7K500 which now occupies the top performance spot for desktop drives, the Seagate Barracuda 7200.9 which is the first shipping Serial ATA-II drive, the Seagate NL35 for backup servers and other "nearline" storage, and the Western Digital WD4000YR, which interestingly is actually based on their famous (and expensive) Raptor unit." Hitachi's SATA-II drive was also recently reviewed by BigBruin in case you missed it.

Comments

Bah

[Anonymous Coward]

I want a solid state drive; sick of mechanical breakdowns and especially the noise.

Re:Bah

[Private Taco]

If it was solid state, would it still be a "drive"?

Re:Bah

[Anonymous Coward]

Then, visit silentpcreview.com

Re:Bah

[Jeff DeMaagd]

I want a solid state drive; sick of mechanical breakdowns and especially the noise.

If you want to pay 100x more per GB of storage, go right on ahead, I won't stop you, but I won't follow your trail either until the cost difference is a lot lower. Even with buying a second set of drives for off-line backup, mechanical drives are still a far better deal for mass storage.

Frankly, I've had not much of either failures or noise in the past five years, unless you are talking 8+ year old drives, but by then they are too small or slow to be useful anyway.

You could simply switch to a quieter drive, like some Samsungs or Seagates, any good drive review site should show relevant noise measurements. I've found some treatments to make very quiet even quieter by gluing sheets of accoustical foam to various parts inside a computer case that might resonate or reflect sound.

Re:Bah

[dtfinch]

At least it wouldn't be a "disk", but I see "ramdisk" and "ramdrive" used interchangeably all the time.

Re:Bah

[evilviper]

I want a solid state drive; sick of mechanical breakdowns

You're sick of mechanical breakdowns, so you want to replace it with something far less reliable, that will frequently have non-mechanical breakdowns.

and especially the noise.

Whaaa? Hard drives are getting much quieter each generation. My 7200RPM 40GB hard drive is extremely quiet. Even with just one very quiet fan in the system, you still wouldn't hear it (except for the inital spin-up). My 160GB hard drive is slightly quieter still, and yeh I added rubber washers to make it slighter quieter (for HTPC use).

Unless you use a completely fanless computer, or are using very old hard drives, I can't understand how anyone can be complaining about hard drive noise these days. You can also use hdparm's "acoustic management" option, to quiet hard drives even further (at the expense of some performance).

Re:Bah

[Kadin2048]

For the price per MB of solid-state storage, you could make a double- or even quadruple-redundant RAID array, and probably buy yourself a nice set of noise-cancelling headphones...

Frankly I think the noise complaint is probably not going to be a big issue among most people, given that for example there is more ambient noise in my home just from the street outside, and in my office from the air handling / HVAC system, than is ever made by any of my computers' hard drives. You'd have to get that environmental 'noise floor' down very low for HD noise to become an issue. If your home or office is really that quiet, it's impressive. Pretty far from average though, I'd say. Especially since most computers have fairly noisy air-cooling systems anyway.

Unless perhaps there are some extremely noisy drives that I'm just not aware of -- the most recent one I've bought was a 120GB PATA model a year or so ago and I certainly can't hear whether it's spinning or not when my computer's case is closed. The last time I remember thinking a hard disk drive was noisy was in regards to an external one that sat under my Macintosh Classic II, and it held a whopping 20MB. :)

Re:Bah

[Cheetahfeathers]

Seems like you're not very sound sensative. Not everyone has the same range and sensativity of hearing, you know. You might not be bothered by it, you might not even hear it, but some of us can. Total silence is the goal of many people with regards to their computer. Even the quietest drive reviewed here bottoms out at above mid 30s dbs. I personally find 20db to be fairly loud for fans and such in my system.

Re:Bah

[Hal_Porter]

Actually, if you want less heat/noise a 2.5" disk isn't a bad choice.

Flash, as everyone else points out has some serious issues - the price per GB is sky high, it has a lower life than hard disks even with wear levelling, and the write speeds are still much slower than a typical hard disk.

My guess will be that flash will never take over from hard disks - hard disk capacity is growing faster than IC capacity, so the price per GB comparison will get worse with time rather than better.

WTF

[neologee]

why is my hard drive ticking when i go on that site!!

Re:WTF

[Hal_Porter]

Ticking harddrives are about to quit on you.

I'd say it's about to quit on you because it's jealous. You shouldn't oggle other younger drives when your old drive is watching.

Seagate's "nearline" drive

[SuperBanana]

Storage manufacturers have tackled the issue by introducing a new class of device, the "nearline" drive that, when combined with the aforementioned online and offline segments, tiers today's enterprise storage into three distinct levels. By keeping highly-accessed, current information in the traditional domain of high-speed, swift-actuator drives and relegating less-used but still-accessed data to slower, less expensive devices, drive firms aspire to deliver solutions that balance the needs for performance with cost.

So to cut through the jargon crap- in other words, someone finally remembered that RAID means Redundant Array of Inexpensive Disks, and that in most cases, when you've got 5 or more drives in an array, you don't need them to be 15,000 RPM?

Re:Seagate's "nearline" drive

[drsmithy]

So to cut through the jargon crap- in other words, someone finally remembered that RAID means Redundant Array of Inexpensive Disks, and that in most cases, when you've got 5 or more drives in an array, you don't need them to be 15,000 RPM?

RAID improves throughput, but not latency. If you need low latency, you need high-RPM drives and no amount of RAID will help you.

Re:Seagate's "nearline" drive

[Noose For A Neck]

No.

RAID does not lower data access times. If you are running an application with lots of random disk accesses e.g. a database, you generally will not be helped by a RAID array in terms of performance unless you are maxing out disk throughput. As 15,000 RPM disks have lower access times, they are used for these sorts of applications. That is why companies are willing to pay outrageous sums for 15,000 RPM disks.

Re:Seagate's "nearline" drive

[__aaxwdb6741]

Wether the I in RAID stands for Inexpensive, Independent, or Irritating, is not a fact - It's a technology, just like SATA is.

Anyhow. I am a Seagate purist. I have two Seagate Barracuda 80GB drives, and one Seagate Barracuda 7200.7 120GB drive. I'm extremely satisfied, as I have yet to experience a harddrive failure - And these drives have been running since I bought them over about a year ago, for 24/7 (My system's uptime is 96,3%).
The review of the 7200.9 drive which was linked to in the article, disappointed me quite a bit. I was hoping to gain a Cool Factor point or two by having Seagate drives.
Nevermind, I'm still a Seagate fanboy.

Re:Seagate's "nearline" drive

[Seumas]

As a recent Raptor purchaser, I can vouch that a RAID of raptors certainly would never fall into "inexpensive" land. 74gb for $170. Ouch.

But they're pretty fucking slick.

Re:Seagate's "nearline" drive

[Alien Being]

"RAID improves throughput, but not latency."

Argh! This myth needs to end.

The only case in which RAID does not improve latency is that of a single tasking system.

The latency that's important for a multitasking system is the time an application has to wait for its data, not the time it takes the disk to process a single request. The benefits vary depending an access patterns, array geometries and RAID level.

Having more drives simply means there's a better chance that some requests can be handled in parallel. Your claim is akin to saying that people won't have to wait longer at the supermarket checkout when only one lane is open.

Re:Seagate's "nearline" drive

[Anonymous Coward]

Hey, Seagate Fanboy! I'm a Quantum SCSI Fanboy! Why are you so thrilled with drive life "over about a year ago"?

Quantums are the 4-barreled Holleys of computing!

Seeya Seagate Fanboy!

Luv,

Quantum SCSI Fanboy

Re:Seagate's "nearline" drive

[Just Some Guy]

RAID does not lower data access times.

That's not strictly true. If you're running a mirror setup, then the odds of one of the drives being available to serve a random read request is better than in a single-drive setup. Put another way, if you have four drives and four processes reading from them, you could theoretically have a single drive dedicated to each process. The average latency would indeed by significantly lower than in a non-RAID system.

I know what you meant, and you were correct within that context, but there are several ways to look at the issue.

Re:Seagate's "nearline" drive

[drsmithy]

Argh! This myth needs to end.

It's not a myth.

It's not an absolute, either, I'll grant you - but it is an excellent rule of thumb.

The only case in which RAID does not improve latency is that of a single tasking system.

This is not correct. RAID *might* improve your latency if its purpose is very specific, the setup can be carefully tuned for the access patterns and the physical placement of data on the disks is predictable, but in general it won't.

The latency that's important for a multitasking system is the time an application has to wait for its data, not the time it takes the disk to process a single request.

I'm confused. How isn't the time a disk takes to process requests directly related to how quickly the data can get to the application, in the general case ?

Having more drives simply means there's a better chance that some requests can be handled in parallel.

Certainly, but the chances of it happening are very low. A higher RPM drive will give immediate, predictable and consistent improvements in access times. A RAID array *might*, some of the time, if you're lucky and the planets are correctly aligned - but on average it will actually make latency worse.

Your claim is akin to saying that people won't have to wait longer at the supermarket checkout when only one lane is open.

Your analogy sucks. Not only is the scenario of people being served at checkouts talking about completely independent operations, but that independence also allows for performance hotspots (ie: longer queues in a particular aisle) to be avoided. Accesses to a RAID array exhibit neither of these characteristics.

Re:Seagate's "nearline" drive

[Crag]

More drives means more heads, means the average distance a head has to travel to read data is shorter, means lower latency.

Are there situations where RAID won't help you? Of course.

Is RAID a silver bullet? Of course not.

Do five drives in a 25% full or less RAID4 (parity not striped) find the data you're looking for faster than one of those drives? Usually.

Does the raid slow down as it fills up? If the heads have to sweep the whole platter to find data, yes.

It is true that for individual operations RAID5 does not improve latency because all of the drives have to be read to verify parity, but this is not true for every RAID algorithm, and few RAID5s are actually used in this way anyway. As several other posters pointed out, improving throughput has the effect of improving overall latency by keeping the I/O queue moving faster. As long as the RAID is always busy, more heads (for the same amount of storage) will improve total latency and throughput.

Re:Seagate's "nearline" drive

[Alien Being]

"Your analogy sucks. Not only is the scenario of people being served at checkouts talking about completely independent operations, but that independence also allows for performance hotspots (ie: longer queues in a particular aisle) to be avoided. Accesses to a RAID array exhibit neither of these characteristics."

The analogy works well. You just seem to be overlooking the obvious. Take mirrored RAID as an example. Assume the system needs to read 100 separate chunks of data. With a single disk, it must wait for 100 seeks (100 people in a single checkout lane), With two disks (lanes), the reads (checkouts) happen in parallel, so the system only waits for 50 seeks.

That's why it IS a myth that RAID doesn't improve latency. It won't improve the best case, but it will improve the average. It's sysadmin 101 that you add spindles when the transaction load gets high. Whether you spread the load at the filesystem level or at the block level doesn't matter, as long as it allows two disks to work at the same time.

Re:Seagate's "nearline" drive

[Anonymous Coward]

"If you need low latency, you need high-RPM drives and no amount of RAID will help you" - by drsmithy (35869) on Saturday November 05, @06:58PM

Oh, really? You're jumping the gun, & overlooking the Solid-State Drive world...

For instance - Do you know that CENATEK RocketDrives can be spanned into a single large 16gb unit (4x4gb each), & their access times &/or latency is FAR lower than ANYTHING even the UltraScSi 15k rpm world (even in RAID) can produce??

Look into them:

http://www.cenatek.com/

I could see using this for DB work, especially temp tables work (if your actual table are WAY too large for 16gb that is, & some are), &/or if your tables don't exceed near to 16gb size.

(And, yes, they have backing powersupplies, which you can additionally supplement with UPS's)

APK

P.S.=> Are there larger models than 4gb each out there in the Solid State drive world? Probably!

I just am not currently aware of them, but they probably are out there... & I do not look into that possibility myself, but consider the fact you may be overlooking them when you made your statement here above is all... apk

Re:Seagate's "nearline" drive

[drsmithy]

The analogy works well. You just seem to be overlooking the obvious. Take mirrored RAID as an example. Assume the system needs to read 100 separate chunks of data. With a single disk, it must wait for 100 seeks (100 people in a single checkout lane), With two disks (lanes), the reads (checkouts) happen in parallel, so the system only waits for 50 seeks.

This is an absolute ideal scenario, and the only one (concerning a RAID1, at least) in which your claim is correct. I hope I don't have to tell you how unrealistic that is to be drawing general conclusions from.

Your analogy doesn't work very well because, as I said, all the "checkouts" can operate independently, and a bottleneck on one does not affect the others. This is not at all like accesses to a RAID array, except in absolutely ideal situations (the kinds you read about in magazine reviews and sales brochures).

That's why it IS a myth that RAID doesn't improve latency.

On average, in general, it doesn't. It does in some cases - in particular where the files are large and accesses contiguous (such that the throughput "overwhelms" the seek time). However, most environments will benefit far more from a smaller number of lower latency drives, than a larger number of higher latency drives, assuming performance is the primary criteria.

It won't improve the best case, but it will improve the average.

No, it won't. It might improve the best case, but not the average, not in general. Indeed, the "best case" (nicely symmetrical, contiguous, parallelisable disk reads) are about all it will improve. In a RAID1 (to use your example) every operation except a disk read can only ever be - even in absolutely ideal, perfect conditions - as fast as a single drive. In actual usage they will be worse.

It's sysadmin 101 that you add spindles when the transaction load gets high.

That's a pretty dramatic oversimplification. It's not difficult at all to come up with scenarios where replacing drives with ones that have lower latency will produce better results than just adding more spindles to an array will.

Added to that, there's a substantial difference between just "adding spindles" and "adding spindles to existing arrays" which is being glossed over.

Whether you spread the load at the filesystem level or at the block level doesn't matter, as long as it allows two disks to work at the same time.

This will not necessarily guarantee an improvement in performance.

In general, the more spindles you add to a RAID array, the worse the latency gets and the better the throughput gets (in particular, with "cheap IDE RAID" style arrays that are obviously being referred to, that usually don't have synchronised spindles). If the access patterns are such that the increase in throughput overwhelms the increase in latency, then overall performance will improve (and, of course, vice versa). However, in my experience that's not a common case - particularly in heavily loaded multitasking server environments. You're better of getting lower latency disks, or adding more separate disks and/or arrays, rather than adding more disks to an existing array.

Of course, the different levels of RAID each have different performance dynamics as well, so it's hard to talk specifically about whatever examples you might be thinking of - but the rule of thumb that RAID doesn't really help with latency applies to all of them, just in relatively different amounts.

From what you've said thus far, I suspect you're using experience with workstation or desktop setups with RAID1 or RAID0 arrays to base your opinions on. This isn't unreasonable in and of itself, but these sort of scenarios will really show a disproportionate impression of performance improvement from higher throughput, compared to heavily loaded and multitasking server scenarios, because the disk activity is far more likely to be linear and contiguous, and because the performance is only averaged across a single user. You can't really draw conclusions about server performance based on workstation usage.

Re:Seagate's "nearline" drive

[Anonymous Luddite]

cost is relative.

Anyone remember when a single gig drive cost 4 figures and was the size of your lunchbox? I do. Maybe I'm stuck in a different decade, but I think Raptors are pretty cheap and 74GB quite big...

Re:Seagate's "nearline" drive

[Alien Being]

"the rule of thumb that RAID doesn't really help with latency applies to all of them"

All of them? You started off by saying that I gave "the one" example of RAID reducing the overall latency of multiple reads. Well, it's not the only one. RAID-0 will give the same result unless the data is pathological somehow, e.g. it clashes with some other interleaving scheme.

I understand what you're saying about the independence of checkout lanes, but it's the same as having a fairly uniform distribution of read requests (across the range of blocks in the volume). A hotspot in the data is similar to having just one checkout lane that accepts checks. Everyone in the store who is paying by check will experience a higher latency. But anyone who has cash will still benefit by the extra lanes.

More spindles == faster service. Naturally, quick drives are also a critical factor, but it's not always possible to upgrade to a quicker drive.

I suspect that the reason so many people think that RAID doesn't reduce latency is that they're using braindead controllers. I've seen plenty of them that are too stupid to even load balance across mirrors, the simplest case.

Re:Seagate's "nearline" drive

[Anonymous Coward]

The type of applications that require my computer to be in an active state all the time, utilize a lot of harddisk operations. The sequence, frequence and type of the operations are quite damaging. I've read an article that this application even "Destroys your harddisk".

Re:Seagate's "nearline" drive

[Anonymous Coward]

Only true when the random accesses are strictly serial, as soon as you have more than 1 queued RAID can drive down the access times (as perceived by the system, ie. the only one which matters).

Re:Seagate's "nearline" drive

[drsmithy]

All of them? You started off by saying that I gave "the one" example of RAID reducing the overall latency of multiple reads.

I pointed out that you picked the one aspect where latency *might* be improved. There's typically a lot more to disk access than perfectly parallelisable disk reads.

I'm talking about *average* performance in the *general* case. On *average*, in *general*, RAID will increase latency. There are certainly examples of specific situations where this is not true, but that doesn't change the *general* case. Particularly when those examples only actually manifest in ideal conditions.

Well, it's not the only one. RAID-0 will give the same result unless the data is pathological somehow, e.g. it clashes with some other interleaving scheme.

Which is quite likely. There aren't many scenarios where it's possible to specifically tune the stripe size to the file size, nor control the physical placement on disk of logical files.

For RAID0 to give the same result, you would need 2 simultaneous reads <=$STRIPE_SIZE, where the data was on different disks. Again, not likely to be a common case.

I understand what you're saying about the independence of checkout lanes, but it's the same as having a fairly uniform distribution of read requests (across the range of blocks in the volume).

No, I don't think you do. The reasons the checkout lanes are "independent", is because the progress of any arbitrary customer through any arbitrary checkout lane is completely self contained. No matter how fast or slow one lane may be progressing, they have no affect on the "processing speed" of any other lane. This does *not* apply with most RAID arrays as typically a logical array transaction will have to touch multiple physical drives. That is, your overall access time to the array, per operation, is dependent on the slowest drive's (involved in the transaction) individual seek time, because the transaction as a whole cannot complete until the slowest drive has. Hopefully you can see that this cannot be better than any individual drive's seek time and, on average, will be worse.

More spindles == faster service.

This is not necessarily true. Again, if the IO patterns are such that seek times dominate throughput, then more spindles will probably not help and, indeed, might even make the situation worse.

The general equation is: transaction_time = (request_size/throughput) * device latency. If throughput can easily dominate (typically, a larger request_size), more drives will help. If it can't, they won't.

RAID really shines with non-small, sequential reads and writes - because these maximise its primary benefit of throughput - this is where your statement above holds. However, when you get into lots of small, random disk requests, the latency of individual drives becomes more the limiting factor, as the transactions aren't big enough to really benefit from the increased throughput. For example, on a usenet server or squid server, you will probably get better performance out of disk subsystem using x low latency drives (be they accessed individually or as a RAID array), than you will from one using 2x drives that have higher latencies, even thought the second scenario would probably have more than twice the throughput in a sustained read.

I suspect that the reason so many people think that RAID doesn't reduce latency is that they're using braindead controllers. I've seen plenty of them that are too stupid to even load balance across mirrors, the simplest case.

The problem is you are taking your hypothetical, best-case scenario and treating it as the general case. This is not a good thing to do. Most of the time your x-disk RAID array isn't going to be able to process x simultaneous disk operations or, indeed, anything even approaching that level of efficiency.

No controller can change the laws of physics. If your disk operation has to touch X disks i

Re:Seagate's "nearline" drive

[Hal_Porter]

You mean BitTorrent?

You can use a client that can pre allocate files, like BitTornado. I can leave it running with half a dozen sessions active and I can't hear the drive seeking.

I'm not 100% sure, but I suspect that running drives warm and constantly seeking reduces their life. You can check the drive temp by reading the SMART attributes. One of these is typically temperature. Make sure that it's comfortably below the max temp in the datasheet. If it isn't, cool the drive with a fan.

Re:Seagate's "nearline" drive

[Anonymous Coward]

The Rocketdrives are rather poor; they've been out-innovated by Gigabyte. Look into Texas Memory Systems stuff. ... and get your wallet out, you will need it.

Re:Seagate's "nearline" drive

[Seumas]

75gb is not that big in an age where a new videogame is more than 6gb. When you've become used to buying several gigabytes for a dollar, spending several dollars for a gigabyte becomes absurd.

But I do still remember the first time I bought a 1gb drive for $200. Hell, I remember my first 20mb hard drive, which was about the same price. ;)

Re:Seagate's "nearline" drive

[Alien Being]

"There's typically a lot more to disk access than perfectly parallelisable disk reads."

"The problem is you are taking your hypothetical, best-case scenario and treating it as the general case."

I did not show the best case nor did I treat it as the general case. I used one of the simplest cases to illustrate my point.

"So, I suppose in a manner of speaking "latency" has been reduced (in that you're waiting less)"

You're almost there.

"but it's not because the latency of the disk operations has improved, it's because the throughut has improved."

No, it's a latency issue. It has nothing to do with throughput, by which I assume you mean transfer rate. Consider the case where each read is for a single byte, or in my analogy, each shopper has a single item. Throughput doesn't matter, it's latency.

You're correct in saying that the more knowledge the RAID controller ( whether it be hardware or software ) has about the data, the more it can optimize performance, but it's only in the patholgical cases where there will be no improvement.

You're also right to say that quicker disks are a win. But even the fastest disk in the world can be easily swamped by a large number of requests.

more spindles == faster service

Re:Seagate's "nearline" drive

[Anonymous Luddite]

>> a new videogame is more than 6gb

If they're that big, I suppose you're right. I haven't bought a game in a few years. Don't play too much, don't download movies or anything so I actually have unpartitioned space on most of my drives.

Guess my screen name is appropriate.

Re:Seagate's "nearline" drive

[drsmithy]

I did not show the best case nor did I treat it as the general case. I used one of the simplest cases to illustrate my point.

Huh ? Perfectly balanced, parallelisable IO (as per your RAID1 and "checkout" examples) is about as "best case" as you can get when talking about RAID IO.

And you *are* treating it as a general case by saying such IO will be the rule, rather than the exception.

You're almost there.

Well, I'm starting to understand what you're trying to argue. That doesn't make it right with regards to what I'm saying, however.

No, it's a latency issue. It has nothing to do with throughput, by which I assume you mean transfer rate. Consider the case where each read is for a single byte, or in my analogy, each shopper has a single item. Throughput doesn't matter, it's latency.

Look, your shopping analogy *does not work* in the general case because it is based on the assumption that each checkout (ie: disk) can operate independently of, and simultaneously with, all the others. You need an ideal scenario for this to be true, and these are rare. Typically an IO to a RAID will have to touch multiple disks, and will also not access the disks in an equitable fashion. Repeating carefully selected best case scenarios won't change the general case.

There are only two cases where a RAID array can even have latency equivalent to a single drive, and that is when a) the IO only touches a single component in the array, or b) the array components are all perfectly synced and the IO operations to the individual drives are processed and complete perfectly parallel - and when you're talking about a single drive with a lower latency than any of the array components, neither of them are possible.

You're correct in saying that the more knowledge the RAID controller ( whether it be hardware or software ) has about the data, the more it can optimize performance, but it's only in the patholgical cases where there will be no improvement.

An operation that touches multiple drives in a RAID array *cannot* have better latency than a single drive operation. No amount of controller of software intelligence can change physics and maths. It may have higher throughput, but whether or not that makes the overall operation complete quicker depends on the request size. A single drive with a 4ms latency but only 75MB/s transfer rate will complete a 300kb IO quicker than a typical 4 disk RAID0 with 170MB/s of throughput and a 10ms latency, even if we assume a perfect best case scenario for the RAID array.

You're also right to say that quicker disks are a win. But even the fastest disk in the world can be easily swamped by a large number of requests.

I never suggested a single drive will always offer better overall performance than a RAID array. Merely that it will have better latency.

more spindles == faster service

This is not always true, and no amount of repetition on your part will change that.

There are two distinct parts of any disk IO operation. How quickly the device can be ready to transfer the data (the latency) and how quickly the data comes off the device (the throughput). Even in an absolutely ideal situation, a RAID array *cannot* have better latency than an individual disk (assuming the disk is of equal or better performance than the RAID components). It might be equal, but it can't be any faster (and will usually be slower). OTOH, an array will have higher throughput. So, if your IO operations are throughput-bound (generally, large and sequential) the RAID will be faster. However, if they're latency bound (generally, small and random) the individual disk (or RAID of fewer, lower latency disks) will be faster. Only when the throughput advantage starts to overwhelm the latency advantage, will overall performance increase (which is what I think you mean when you say latency).

These two principles follow through into RAID arrays themselves. The more spindles you add, the worse lat

Re:Seagate's "nearline" drive

[Alien Being]

"I never suggested a single drive will always offer better overall performance than a RAID array. Merely that it will have better latency."

I wasn't trying to imply that you had made that suggestion. Just to be clear, this whole discussion has been about latencies, not transfer rates.

"Huh ? Perfectly balanced, parallelisable IO (as per your RAID1 and "checkout" examples) is about as "best case" as you can get when talking about RAID IO."

With RAID-1 it doesn't matter that the requests are balanced. a two-way mirror gives a 2x improvement in average latency. Make it a 3-way mirror, and you'll have a 3x improvement.

With RAID-0, the balance of requests does matter. Assume two disks, one containing even block numbers, the other containing odds. If the requests are pathological, i.e. all odds, the improvement will be 0. But if *any* of the requests are even, then there's an improvement.

I only jumped into this thread because you were modded informative for saying:

"RAID improves throughput, but not latency. If you need low latency, you need high-RPM drives and no amount of RAID will help you."

High-RPM (and fast head servos) are important, I never said otherwise. but improvements in those areas come incrementally. Many systems already have the fastest disks available, yet the demand for faster disk service persists and RAID satisfies that demand.

Wikipedia backs up what I have said. Please read what it says.

Re:Seagate's "nearline" drive

[Anonymous Coward]

"The Rocketdrives are rather poor" - by Anonymous Coward on Sunday November 06, @12:15PM

I dunno, I like mine alot! I use 1/2 as my pagefile (1st partition) & 1/2 as my temp/swap area for apps & the OS, @ least for my home system.

I got mine on a HEAVY discount, I wrote up a review for them is why. Only downside I noted? COST!

http://www.cenatek.com/

" An Independent User's Review of CENATEK's Rocket Drive"

(Mine was along with many others, including Windows IT Pro mag, & they knew who I was, since I wrote a competing ramdrive product YEARS ago based off MS' DDK driver template... nice part? I got "front page of their site" status, even over Windows IT Pro mag's review & others from this page:)

http://www.cenatek.com/press.cfm

The nicest part, other than that front page status on their site & getting mine @ 1/3rd the std. cost? Was I got a great product & love having it! Nothing touches it for speed of access & putting my pagefile.sys into RAM, literally?

FLIES!

(As well as temp ops of ALL kinds from %System/User Environment vars% in the System Control Panel applet section on performance tuning in Windows NT-based OS flying too!)

Other than that high cost for typical buyers (which I was not subject to for writing that review), I found nothing but GOOD things to say about it.

"they've been out-innovated by Gigabyte" - by Anonymous Coward on Sunday November 06, @12:15PM

BUT, not by even 15k rpm diskdrives from the UltraScSi world & my init. point here in reply, point-blank!

About being "out-innovated"? It happens - natural fact in this, or any, industry. I've seen buildings in various companies where they pull apart a competitors work (reverse engineering) & steal its ideas, & do it better/faster/cheaper... not TOO hard to do, once you have a template to work on (the other guys work) & to start from!

It happens in software like mad (I am a software engineer for work typically the last decade/decade ++ now) & also in the world of hardware.

Anyhow - on the init. topic? Nothing can touch solid-state disks for access times & latency... nothing from the std. world of mechanical diskdrives (even the fastest 15k rpm diskdrives from the UltraScSi world).

Point-blank/fact.

APK

Re:Seagate's "nearline" drive

[drsmithy]

I wasn't trying to imply that you had made that suggestion. Just to be clear, this whole discussion has been about latencies, not transfer rates.

The problem is you don't appear to be using the term "latency" in a meaningful or consistent fashion.

With RAID-1 it doesn't matter that the requests are balanced. a two-way mirror gives a 2x improvement in average latency. Make it a 3-way mirror, and you'll have a 3x improvement.

False.

With RAID-0, the balance of requests does matter. Assume two disks, one containing even block numbers, the other containing odds. If the requests are pathological, i.e. all odds, the improvement will be 0. But if *any* of the requests are even, then there's an improvement.

In transfer rate, yes. In latency, no. In overall performance, maybe.

Wikipedia backs up what I have said. Please read what it says.

That doesn't make it correct. Indeed, whoever wrote that Wikipedia article has the same problems with terminology and assumptions you do.

I'll make one last attempt to explain why RAID will not improve latency, and will generally make it worse. If you can't see the physics and maths that make it so, there's not going to be much I can do to change your mind.

Consider the scenario where you have a three-disk RAID0, with a 64kb chunk size. A request for 140kb is made, and conveniently enough, is contained in chunks on all three disks. So they are able to operate in parallel and provide maximum benefit.

Disk 0 seeks to the appropriate place and takes the average seek time of 8ms.

Simultaneously, Disk 1 seeks to the appropriate place, however, since the necessary block is further away from the head, its seek time is 11ms.

Disk 2 is lucky, and the necessary data is quite close to the read head. It's seek time is only 4ms.

The disk return the the chunks, which are appropriately assembled by the RAID layer and passed back up to the OS (since the requests are so small, the transfer times in this example are negligible - this is the sort of disk operation where a RAID array's performance suffers).

The latency for this operation was 11ms (plus a tiny amount for transfer). The time taken by the slowest seek. Why ? Because the operation could not complete until both chunks of the stripe had been retrieved, even though two of the three disks involved only took 4ms and 8ms.

Now, if we take a more ideal view and say that all three disks only take the average seek time (8ms) to return their chunks. The latency for the operation is *still* 8ms, because it cannot complete until all three disks have returned their chunks. It will most certainly *not* be <3ms, which is the result you would claim was possible (8/3).

The RAID array can never have have a latency lower than that of the component devices. Obviously, since that value can only ever even match the component devices in ideal conditions (perfectly parallelised operations), it will be higher on average than the component devices' average. This is why the average latency of a RAID array will be higher than the average latency of a single drive, even if that drive is identical to the RAID component drives.

(This is the same logic that dictates a RAID1 cannot have write performance better than an individual disk.)

Also, this example is talking about a RAID0, which is about as best-case as you can get in terms of overall RAID performance. Other RAID levels that offer no advantage - or even a penalty - for writes (particularly random writes) like RAID1 and RAID5, only swing the latency advantage further towards the individual disk case.

Here is another way to think about it. By your reasoning, a RAID1 made up of eight disks with an 8ms latency would have an average latency of 1ms. Do you seriously believe if you benchmark an eight-disk RAID1 you'll get an average seek time of 1ms ?

Re:Seagate's "nearline" drive

[Alien Being]

"The problem is you don't appear to be using the term "latency" in a meaningful or consistent fashion."

Latency is the time the program has to wait for a minimal amount of data, i.e. a one-byte request. It includes the time it takes to dispatch the command, the time it takes the head to position itself and settle at the correct cylinder, the time it takes for the platter to rotate to the beginning of the target sectors and whatever bus latencies are involved. That's for a single request. The thing that you seem to be missing is that the latency also includes the time waiting for earlier requests to be serviced (even without considering xfer time).

"That doesn't make it correct. Indeed, whoever wrote that Wikipedia article has the same problems with terminology and assumptions you do."

It's a better source than any you have bothered to find.

"Consider the scenario where you have a three-disk RAID0, with a 64kb chunk size. A request for 140kb is made, and conveniently enough, is contained in chunks on all three disks. So they are able to operate in parallel and provide maximum benefit."

In the cases where the single disk can fetch the 140k with a single seek, it will be faster on average than the 3-way raid-0. But a request that requires >1 seek on the single disk will have a shorter latency when distributed across >1 disks. Three requests for chunks that only require a single seek on a single disk can be handled simultaneously, each with the same average latency as the single disk.

"Here is another way to think about it. By your reasoning, a RAID1 made up of eight disks with an 8ms latency would have an average latency of 1ms. Do you seriously believe if you benchmark an eight-disk RAID1 you'll get an average seek time of 1ms ?"

The mistake you're making is not looking at the workload as a whole.

Take 8 random requests and submit them to single disk. I'd expect to wait (on average) 64ms for them all to complete. Now submit them to an 8-way RAID-1. I wouldn't expect any of them to complete in 1ms. That's not what I've been saying. But I would expect them to complete in 8ms. So instead of 64ms/8 requests, I'd have 8ms/8 requests.

It's more about statistics than physics.

Your point about writes versus reads is well understood. Parity RAID is very slow to write, and mirrors will also increase the average latency. RAID-0 however provides the same speedup on writes as it does on reads.

Testbed

[msbsod]

Why do the folks at storagereview put such nice harddisks into testbeds with Microsoft Windows? At least they mention Anti-virus software. Safer hard drive, always!

Re:Testbed

[Noose For A Neck]

Because most people with computers use Windows and there has been a lot of disk benchmarking software developed for Windows.

Besides, a lot of their forum regulars are fanatical Windows supporters, to the point of claiming "...in smp scaling windows is better than linux, and linux is better than freebsd. for vms windows is better than freebsd, and freebsd is better than linux.", so I'd take what they say with a grain of salt.

Re:Testbed

[NineNine]

Why do the folks at storagereview put such nice harddisks into testbeds with Microsoft Windows

Because that's what most people use, maybe? Because that's what most benchmarking tools run on, maybe? If anything, I'd wonder why somebody would do benchmarks on something other than Windows.

Re:Testbed

[Cal Paterson]

Because that's what most people use, maybe?

Hmmmm, I would've thought that most of the people interested in harddrive reviews would be small businesses, home website owners etc, and I assume that most of these guys are using some form of Unix. Incorrect?

Re:Testbed

[Tim Browse]

Incorrect?

I'd say so, yes.

Re:Testbed

[msbsod]

Calm down. It was a joke. Perhaps I should have written my comment in LaTeX.

Re:Testbed

[ocelotbob]

Because people with real work to do use operating systems other than windows, perhaps? Windows is fine for games and serving dodgy content, but those with real work to do prefer much more stable, proven operating systems.

Re:Testbed

[ZenShadow]

Windows is only as unstable as the administrator that's managing it.

That said, many Windows administrators are notoriously unstable due to the low barrier to entry. That's one of the main reasons I prefer Linux on my servers for most applications. :-)

--S

Re:Testbed

[Tony Hoyle]

Administrators can be quite unstable.. especially if you're the 1000th user to ask "has the network crashed?"

Re:Testbed

[Reziac]

For the consumer market, I agree, but for the enterprise market, I'd think you'd want to at least do a token test on some species of Unix.

But likely the results would be proportionately the same, even if the absolute performance is different.

Performance vs Noise

[Munta]

While high performance drives may be important, I feel that many drive manufacturers are forgetting that noise is a big issue. When you have 3 PC plus running at home high performance drives will just add to already high number of decibels that we have to suffer. Keeping my hearing is probably more important to saving a few milliseconds for each drive seek.

Re:Performance vs Noise

[CyricZ]

Have you invested in a proper server closet? Indeed, many people build them in the basement of their home. You can use various dumb terminal systems to access your computers, even desktop ones. You get the benefits of a desktop under your desk, but without the noise and the increased room temperature.

I know several Scottish folk who even use the chilly winter air to help cool their systems. That may not be an option for you, but if it is, go for it.

Re:Performance vs Noise

[Noose For A Neck]

If that's one of your needs, Storagereview also measures the noise output of each drive in their reviews, along with heat output.

Re:Performance vs Noise

[NineNine]

If you think that a hard drive (or any number of hard drives) for that matter could contribute to hearing problems, then I'm assuming that you don't leave your house? Hell, a fart is louder than even the loudest hard drive. I can't imagine there are very many people who are willing to pay extra for super silent hard drives.

Re:Performance vs Noise

[Lucractius]

many people in other parts of the world lack these "basements" you speak of. where can i buy one for myself??? /sarcasm

and the other factor is space. not everyones got room to dedicate to a closet. hell i need to find a Rack for a 6U server i have and you have NO idea how hard that is when you actualy try... it always more expensive when you actualy try it.

Re:Performance vs Noise

[CyricZ]

If you don't have a basement, build a server shelter in your garage. Perhaps even your attic, if the temperature remains low enough.

I know some other people who live in apartments, and they have put their servers/desktops under their kitchen counters. Now, most people have kitchens, even those living in apartments.

And if you can't find a suitable rack, build one. Get some wood, a saw, a few nails, a hammer, and you could be done within an hour.

Re:Performance vs Noise

[Munta]

Well, when you work from home, you could assume I don't leave the house often.

If someone was to stay in a room for 16 hours a day with someone who farts constantly - you may not go deaf, but you would probably die!

Re:Performance vs Noise

[wierdling]

With the price of heating oil, my computers produce the only heating that I have.

Re:Performance vs Noise

[Lucractius]

Obviously you dont own a 6U Alphaserver 2100 RM

it weighs 54 kilos standard, the average wood DIY rack aint gonna cut it for this baby.

Re:Performance vs Noise

[CyricZ]

I administered several of them at a previous job. They're not your typical desktop PC, but they're certainly not 1970s mainframes either.

You obviously do not understand the strength of wood. A properly built server rack can easily handle a 54 kg system. Even the pre-fabricated wooden racks from your typical hardware store are more than sufficient. You can reinforce one of those, if you really feel it to be necessary.

Don't forget that houses are often built from wood. It's a very versatile and strong construction material.

Re:Performance vs Noise

[Lucractius]

Im exceedinly cheep and far too proud of having obtained this large machine in the remote location i live. I want a little 8U Soundproofed glass rack for it ok... you caught me :P

Re:Performance vs Noise

[Jeff DeMaagd]

Keeping my hearing is probably more important to saving a few milliseconds for each drive seek.

That's pretty sad hyperbole. The loudest drive referenced in the 7K500 review is 45dB at 3mm, and that's the WD4000YR, and Western Digital seems to be known for the loudest drives.

In order to cause deafness, I think you'd need to be exposed to 90dB for a long duration, meaning you'd need to be near at least 15 of those drives, assuming you don't have any enclosures or accoustical treatments between you and the drives.

Check the reviews at a good reputable site, Storage Review has bothered to measure accoustical noise and buy the quietest ones.

Re:Performance vs Noise

[ZenShadow]

I can't imagine there are very many people who are willing to pay extra for super silent hard drives.

People would probably pay a lot of money for the guaranteed ability to fart super silently, however. :-)

--S

Re:Performance vs Noise

[Diag]

If you think that a hard drive (or any number of hard drives) for that matter could contribute to hearing problems, then I'm assuming that you don't leave your house? Hell, a fart is louder than even the loudest hard drive.

I just spent the last 2 weeks working in a computer room wearing earplugs due to OH&S regulations, because of the excessive noise in the room being generated by several high end disk subsystems.

5000 people farting simultaneously and continuously in an enclosed space could get a bit noisy (and smelly).

Re:Performance vs Noise

[Anonymous Coward]

Nilesh - is that you?

Lovingly, Paul.

Re:Performance vs Noise

[Yartrebo]

I couldn't agree more. A well-designed shelf using 2x6s should support several TONS of weight per plank. Even standard shelving using much thinner wood and using glue (and maybe screws) instead of quarter inch bolts as fasteners will probably suffice for such a light load as yours.

And yes, as the above poster said, most houses are built of wood. Pretty much wherever corrosion and weight/bulk are not terribly large issues, wood is the material of choice because it's cheap, easy to work with, and more enviormentally friendly than the alternatives (metal, plastic, stone, concrete).

When will we see RAM drives

[ylikone]

... that are as big as hard drives with spinning platters? And when will they be re-writable as many times as hard drives with spinning platters?

I really REALLY want a dependable, long-lasting, fast and ample-capacity RAM drive. No more spinning platters please.

Re:When will we see RAM drives

[Anonymous Coward]

That would be cool, except for losing all your data every time the power goes out (or your computer crashes).

Re:When will we see RAM drives

[CyricZ]

RAM does not imply that it is volatile storage. Remember, RAM stands for Random Access Memory. The emphasis is on being able to access the data without having to step through it sequentially. Thus any technology that allows for random offsets within the storage medium to be accessed directly can be considered RAM.

Re:When will we see RAM drives

[ZenShadow]

On the other hand, the commonly accepted defacto standard is that RAM is volatile. That's why someone came up with the extended acronym NVRAM...

--S

Hard Drive Breakthrough?

[lappy512]

Is there some new hard drive breakthrough now or something? How are so many hard drives coming out?

Re:Hard Drive Breakthrough?

[CyricZ]

Not really. It is just that StorageReview.com has spent a long time setting up their new testbed. The drives aren't exactly new, but all of the reviews are, since they were just written. So the disk developments have been spread out, but the reviews have all been made available at once.

Start by buying today's RAM drives.

[CyricZ]

Then you'll have to start buying today's RAM drives. Remember, the companies need to get financing from somewhere. Sales is one such source of funds. The more you buy from them now, the more they can invest in coming up with the technology you want.

Re:Start by buying today's RAM drives.

[ylikone]

Hmmm.. well, so far, the best I'm seeing is some 40GB drives that cost around $500. Way too much to pay for way too little in my opinion.

Re:Start by buying today's RAM drives.

[NineNine]

I think that's pretty reasonable. I may actually pick up a few more some mission critical computers running here that can't afford any downtime (retail point of sale systems). $500 is peanuts to have that kind of reliability.

Re:Start by buying today's RAM drives.

[ylikone]

Sorry, that should have read 4GB, not 40GB. Yes, a measly 4 GB!

Re:Start by buying today's RAM drives.

[CyricZ]

Then you'll just have to wait until those with that much to spend go ahead and spend it. Then that will allow for the R&D necessary for larger, cheaper drives to be funded.

Re:Start by buying today's RAM drives.

[Jamesday]

Which are those? That's cheap enough to be interesting for people using databases with transaction logs being flushed after every transaction.

Re:Start by buying today's RAM drives.

[spectre_240sx]

Either that or any of the big hard drive manufacturers could step up to the challenge with their own R&D lab and beat the others to market with something that pretty much every IT guy would love to see. You're acting like this has to come from some brand new company. Face it; many of the companies that produce computer hardware have grown complacent with their place in the industry. Sure, they all come up with bigger drives every once in a while, but with the amount of money all of these companies bring in, they could stand to take a risk now and then.

Re:Start by buying today's RAM drives.

[Diag]

The thing is, it's not new technology. From http://www.findarticles.com/p/articles/mi_m0BRZ/is _9_22/ai_101679012 ...

The first SSD was delivered to the MVS mainframe market in 1978 by StorageTek and sold for $8,800 per megabyte-much cheaper than add-on memory-and had a maximum capacity of 90MB.

(SSD = Solid State Disk)

It had moderate success back then, but then disappeared as mainframe memory and disk costs dropped. Obviously it can be done a lot cheaper today, but in comparison to the cost of disk, it's still hard to justify the cost, particularly in business. And realistically, the home market for this kind of thing is kinda niche.

Noise?

[Billly Gates]

Most places like tomshardware and other benchmarking sites mention that hard drive speed is almost neglibable between teh high end and low end drivers for real desktop apps.

I care about reliability (gone down hill since 2000) and noise. I sense in the rush to devalue pc's into $399 emachines that quality is looked upon last in an effort to cut costs. Isn't there anyone buying anything besides junk anymore? I am not talking about servers either since scsi drives and cards are outrageously expensive.

Re:Noise?

[Billly Gates]

I meant "drives" not "drivers" ... now shoot me

Re:Noise?

[Lucractius]

look at hiatchi laptop drives like the ones out of IBM (now lenovo *shudder*) laptops. You may find all those attirbutes you seek there.

Re:Noise?

[dfjghsk]

look at Seagate.. only manufacturer (that I know of) that has a 5 year warranty.. I have 8 of them, no problems yet. Too bad their latest drives are made in China :(

Re:Noise?

[standsolid]

Sure, I buy apple computers.

Re:Noise?

[LurkerXXX]

Yes, quality died in the late 90's, but it's made a comeback in some places in the last few years. After cheaping out and shortening the lenght of their warrenties to the 1-2 year range of the low end competition, Seagate is back to 5 year warrenties. Several other manufacturers have also followed suite, raising quality and their warrenties.

Size vs. Noise

[fredistheking]

The problem is that the density per disc is currently only about 125GB. So any drive over 125GB will have more than a single platter. The WD4000YR has 5 platters as do the Hitachi 400GB and 500GB. This extra mass is the problem which leads to higher temperatures, more noise and more power consumption.

160GB per platter is just on the horizon and with newer TUMR heads expect that to go up to 250GB per platter in a year. I wouldn't hold my breath for perpendicular recording. It still seems to be a couple years away.

Re:Size vs. Noise

[Snover]

Seagate has 133GB platters that use perpendicular recording.

Re:Size vs. Noise

[Snover]

Oops, I definitely am getting some wires crossed.
Seagate has 80GB platters that use perpendicular recording in their 160GB Momentus 5400.3 drives. Toshiba also has 40GB platters that use perpendicular recording in their 40GB MK4007GAL 1.8" drives.

Re:Size vs. Noise

[bani]

drives with perpendicular recording are already shipping. however iirc they are mainly in 1.8" and 2.5" drives where the cost/performance can be justified.

cache memory

[SuperBanana]

RAID improves throughput, but not latency. If you need low latency, you need high-RPM drives and no amount of RAID will help you.

That's generally what cache is for- on the drive, on the controller, and in the system itself. Even databases are optimized with the "expense" of disk access kept in mind in the optimizer's strategy.

Re:cache memory

[drsmithy]

That's generally what cache is for- on the drive, on the controller, and in the system itself. Even databases are optimized with the "expense" of disk access kept in mind in the optimizer's strategy.

This does not change the fact that if you need low latency, RAID won't help you much, if at all, and is more likely to make the situation worse.

drive industry moving too quickly

[Anonymous Coward]

first 7,200rpm drive released in 1992
first 10,000rpm drive released in 1996
first 15,000rpm drive released in 2000

It's nice to see mainstream drives have exceeded the spindle speeds of 13 year old drives.

oh wait..

SATA-IO

[Anonymous Coward]

Shouldn't these be called SATA-IO drives? http://www.sata-io.org/namingguidelines.asp

RAID NAS reccomendation

[Andrew Tanenbaum]

The Nearline looks good, does anyone have reccomendations on a NAS I can toss a few of them into for GigE RAID5? The ReadyNAS 600 and X-6 look better than the Buffalo TeraStation, but what else is there?

Re:RAID NAS reccomendation

[Anonymous Coward]

Don't get a Buffalo TeraStation.

No rescue facility. Firmware (OS, actually Linux) on one disk. If that goes, it's all dead. Happened to me. Oh, and the RAID? Custom disk format, not rescuable even if you crack the case and lift the drives. Customer service sucks. Quite pissed off.

My suggestion: Build your own, jeez. Cheap Linux box, slap, I dunno, Debian on it or something. Share it with Samba or whatever. That's all those NAS boxes are anyway, it's just they're built a lot crappier than you'll build it. Use software RAID (LVM). You'll save a lot of money, and trouble.

I'm sorry?

[Noose For A Neck]

I wasn't the guy making those claims, nitwit. It was someone on their forums. Perhaps you should brush up on your reading comprehension skills.

Re:I'm sorry?

[CyricZ]

I read your post just fine. I'm just not sure why you felt it necessary to come here and complain about getting banned over there at those forums. If you make outrageous and obviously false claims, such that Windows scales better than Linux and FreeBSD, then you're bound to run into opposition.

Not quite there

[fredistheking]

Right now you can get 2GB flash chips. Put a few of these together and you have enough space for a decent install. The problem is that the performance is not quite there yet. You can expect raw read and right performance aroun 10MB/s on the fastest chips on the market. This is still alot less than the 50MB/s+ you find on desktop drives.

Samsung will do it before anyone else. WD, Seagate and Hitachi don't have any flash business so they are not going to push for it.

Re:Not quite there

[Anonymous Coward]

A sector of Flash ROM goes bad after only a few thousand writes. Unless all your programs hardly ever write to files, it's useless for a computer hard drive.

Re:Not quite there

[timeOday]

I don't see why performance should be a problem with solid state storage. Why can't they just put more controllers on the device, reading more independent banks of memory, until performance is good enough? Or you could do it at a higher level by filling up a USB2 hub with memory sticks and using RAID on them.

Re:Not quite there

[bst82551]

http://www.engadget.com/entry/1234000770065138/

80 MB/s transfer. $1900.

Brian

Re:Not quite there

[fredistheking]

I don't think $1900 means its ready to take over.

Re:Not quite there

[binarybum]

ehh, wouldn't a mobo that could support 10GB of Ram be a better option? Doesn't solve the boot problem, but you could load a decent amount of relevant apps right into ram from the getgo and be whizzing along alot quicker than 80MB/s.

Re:Not quite there

[bronaugh]

Wear levelling can mitigate this to some extent, but fundamentally you still do have this problem. This is why I haven't moved to flash.

Samsung Samsung Samsung

[slaker]

SR makes a habit of forgetting about the other commodity drive manufacturer, Samsung. How much do they forget? Well, at one point at least, an Australian forum member (Tannin, if you know SR's forums) had to send them a drive to review because they couldn't or wouldn't make the effort to get one themselves. Also interesting is that Samsung has no relationship with SR as far as advertising.

Which is a goddamned shame, because they really are genuinely good drives (far better than the for-shit products Maxtor and WD are shoveling out these days), ones I buy in preference to any other vendor's. They've been extremely reliable for me and have a nice mix of performance characteristics.

I'm not a big fan of their self-reporting reliability database, and I can't hazard to guess why they're testing "desktop" performance in their Enterprise-I/O Xeon system... nor why they can't do any testing on *nix. But those are all are reasons why I have become frustrated with SR over the last few years.

I'm just one person. My opinions aren't going to mean shit to anyone here. But then, I'm one guy with around 12TB worth of Hitachi and Samsung drives keeping his apartment warm, so it's not like I don't have a little bit of experience with commodity hard disks.

Re:Samsung Samsung Samsung

[CyricZ]

What do you use your 12 TB of diskspace for? I assume it would be for personal use, considering it's in your apartment. Do you do audio and video editing?

Re:Samsung Samsung Samsung

[Lisandro]

I have an old Samsung 3.24Gb 5400rpm drive (model WU33205A) that i have in a carry case; i used it when i needed to move big files arround, now it stays put. That drive made *countless* trips all arround the city and took a lot of abuse - and there it is, still working. It just won't die on me.

    In the meantime (quite a few years... that drive was from and old P133 system), i had a number of hard disks (Seagates, WDs, Maxtors) die on me regardless of being treated much, much nicer. Maybe i should take the hint and check what Samsung is up to these days regarding HD technology.

Re:Samsung Samsung Samsung

[Killall -9 Bash]

Samsung makes the worst HDDs i've ever seen. There's a reason they're the lowest priced drives on every wholesale list.... Unacceptable DOA% and none of the ones we sold/used at work outlived their(or our) warranty. We stopped ordering them.

Re:Samsung Samsung Samsung

[slaker]

It's actually about 8TB of real space, not counting spares and redundant drives.

I sorta-kinda built a system for managing a staggering amount of multimedia content + limited metadata, around the idea of having a great deal of disk space and lo and behold, I keep adding more.

I've got a pretty big library of stuff I've vidcapped - VHS movies that won't ever be made into DVDS, a lot of porn (it's fair to say that I have a fairly expansive definition of "a lot"), pictures I've snagged from online... It's all for my own amusement. I like collecting data. And I'd like to add that I managed to accumuluate well over half that amount of data before I got a broadband internet connection.

400GB drives+ are a blessing; now I can keep more of my collection of stuff online at any one time.

Re:Samsung Samsung Samsung

[slaker]

I think you need to look at your distributor's shipping methods if you're having those kinds of issues with Samsung. I'll grant that Samsung 6 and 8GB models were less than stellar (if you remember back that far), but the 20GB+ models have been standouts for reliability.

Re:Samsung Samsung Samsung

[Seumas]

I only have 3 TB.

However, it's ALL porn.

Yes, I'm serious.

Yes, I'm single.

Yes, my right hand is very agile and strong.

Shut up.

Re:Samsung Samsung Samsung

[Jeff DeMaagd]

I think it is hard to track. For every given brand of product, it's easy to find someone that's had impeccable reliability, and it's also easy to find someone that's seen horrid reliability with the same product. I really can't say why. I would expect every product to fail eventually, and possibly for different reasons. It is possible that explains why different people get different results, is that their circumstances are somewhat different, be it temperature, humidity, atmospheric pressure, power, handling, shipping and who knows what else.

Re:Samsung Samsung Samsung

[timeOday]

I think it is hard to track. For every given brand of product, it's easy to find someone that's had impeccable reliability, and it's also easy to find someone that's seen horrid reliability with the same product.

I judge hard drive longevity by the warranty. At least you know it will last that long. 1 year does not speak well of a hard drive.

Re:Samsung Samsung Samsung

[Tripster]

Definitely, I have swapped out all my old WD and Maxtors in favour of the 80GB Samsung drives, they were also one the first drive makers to bring back the 3 year warranty.

They are fast little drives and with the fluid bearings are quite a bit quieter than the drives they replaced.

Re:Samsung Samsung Samsung

[darkwhite]

Reliability reports on drives grouped by manufacturer cannot be trusted except when they're extreme low outliers (e.g. IBM's Deathstar). All major brands today are about equally reliable for practical purposes (except when you have severe heat dissipation issues maybe). In 8 years and no less than 15 different drives of 10 different models, I have not had a Maxtor drive die on me.

In general, I find that clean power is much more necessary than a preferred manufacturer. Dirty power from a cheap or overloaded power supply may kill the most reliable drive ever.

Re:Samsung Samsung Samsung

[LurkerXXX]

For every given brand of product, it's easy to find someone that's had impeccable reliability, and it's also easy to find someone that's seen horrid reliability with the same product. I really can't say why.

Easy, it's statistics. Most people you talk to have dealt with a very small sample number of drives, maybe a dozen or so. If they get happen to get failures they think the brand is crap. If they happen not to, the they think the brand is great.

If you want to get a better idea, talk to someone who deals with thousands of drives, like an IT guy from a large company. Then you start to get the view from a statistically meaningful number of drives.

As the other poster said, you can also get a decent idea from the warrenty. If the company (Maxtor) only warrenties their drives for one year, and another (Seagate) warrenties theirs for 5 years, there is probably a reason the one with the longer warrenty is willing to risk returns over a longer period of time. Namely that they won't get that many returned because they are less likely to fail.

Being in charge if several hundred drives at work myself, and talking with IT staffers at larger corps, I can tell you that the correlation seems to fit well.

You also need to keep in mind that even manufacturers who overall make very solid drives (IBM) can make a certain model of drive (Deathstar) that can have a design defect and give the overall brand a worse name than it may deserve. Certain users will have used only a small number of the manufacturers drives, with a high proportion being of the model with the defect, so the brand will see much worse to them than it is overall. High sampling numbers are the key.

Re:Samsung Samsung Samsung

[Anonymous Coward]

If you go back far enough, Samsung drives were ABSOLUTE shit. I was building computers for a living 15 or so years ago when Samsung first started shipping hard drives. I think they were 200 meg units. We got a case of them. They failed for any reason at all. I once mirrored an image onto one, then while installing it into a case, I set it down so the edge was on the handle of a screwdriver. When bumped, that edge fell about 1cm onto the wood countertop (not running). It was totally dead.
The whole case was like that. They worked OK if you could get them into the machine and running, and if you never ever let them experience any kind of shock.

Obviously they've gotten way better. No hard drive manufacturer should be judged by their old generation drives. Back in that day, every manufacturer went through good phases and astoundingly bad phases; we used to buy cases of Seagate drives, only to have them ALL fail (20 of them) before 6 months. Yet today Seagate is probably one of the best manufacturers. WD also was once the best, and once one of the worst manufacturers.

I'm actually a pretty big fan of Samsung in general these days (for all kinds of electronics). I wouldn't hesitate to buy a Samsung drive.

Then again, I have about 1TB of Maxtor drives spinning for between 2 and 4 years with zero issues in my house. I don't have that many hard drive problems. But I don't beat on them and I make sure they have cooling on them.

lol what

[Noose For A Neck]

no text

Re:lol what

[CyricZ]

I think it'd probably be okay if you made another account over there. Just try to keep your claims truthful. Suggesting that FreeBSD and Linux scale worse than Windows is sure to get your account banned, as such a claim is obviously false.

Re:lol what

[Noose For A Neck]

Buddy, I don't know who you think I am, but I am not the forum user "honold", who made exactly the claims that I linked to in my original post. I have neither been banned from their forums nor posted anything untruthful.

Re:lol what

[Anonymous Coward]

Methinks you've just been trolled.

storagereview; one of the few smart review sites

[nostriluu]

Slightly off topic, but Storage Review has always allowed the end user to compare any device to another by selection (eg 5400 rpm maxtor vs 10000 rpm WD) using discreet data fields (eg noise and heat). No other review site I know lets you do this, and its a very useful feature. Very often other review sites will scatter related devices across different non intersecting reviews, and I doubt they bother to break the data down to this level of detail.

Re:storagereview; one of the few smart review site

[Anonymous Coward]

Also, they use a single standard test platform for a long time. I think that helps out a lot.

I'm off Maxtor

[smartin]

Every Maxtor drive that I've bought in the last few years has died. Now buying Seagate due to their 5 year warrentee, raid 1ing them and will see what happens.

Re:I'm off Maxtor

[Anonymous Coward]

I can say the same for Western Digital. In the past 3 years, I've had a 80GB, 120GB, and two 160GBs die on me. And I mean completely dead, with no chance for data recovery unless you shell out a thousand bucks.

To make things worse, their warranty service has been steadily getting worse too. When I advanced RMA'ed one 160GB, they didn't even send me a replacement drive, but I had 30 days to send my broken one or get charged on my credit card. Sending E-Mails didn't help, I got generic responses from people whose names I couldn't even pronounce. Calling them and yelling for a bit sorted things out though.

I've learned from my mistakes.. I'm no longer going to stick with one company only. Hitachi, Seagate, Maxtor, or whatever, I'm just going to mix and match and hope it works.

Re:I'm off Maxtor

[Anonymous Coward]

Every Maxtor drive that I've bought in the last few years has died.

Disc drives are running hot these days. Heat is the enemy of a long service life.

Mount the drive so you can install a small fan in front, such as from an old power supply. Get a quiet fan with low vibration. Blow air across the top and bottom to cool the drive.

You should see a significant increase in reliability.

Re:I'm off Maxtor

[smartin]

Good advice!

Re:I'm off Maxtor

[WuphonsReach]

There are two nice items that I've found for cooling drives.

1) A 3 in 2 drive bay unit. This allows you to place up to (3) 3.5" drives into a pair of 5.25" drive bays. The unit comes with an 80mm fan on the front along with a front filter. Your best bet is to use this as a 2 drive in 2 bay configuration. That gives more airflow to each drive, plus uses an 80mm fan (wich is quieter) rather then (4) 40mm fans (noisy). I think I found my last one at Mwave.com (part #BA21364). If you have a pair of hot drives and a pair of 5.25" bays free, this unit is the perfect match.

I've used one of these 3:2 bay units for a few years now. Nice flexibility for older cases without a dedicated drive bay fan. The downside is that you have a longer cable run (usually) to get up to the 5.25" bays. But this is a non-issue with SCSI or SATA. Took me a while to find another company that carried them, but finally MWave started stocking them.

2) A 4 in 3 drive bay unit. I found these over at CaseMod.com. Same concept, except that it fits up to (4) 3.5" drives into (3) 5.25" bays. The fan is a 120mm fan (even quieter then the 80mm), but it doesn't come with a filter on the front. Ideal for tower cases where you have (4) 5.25" bays and you've put a CD/DVD in the top bay and have the next (3) bays open.

For better cooling, you could stick with only putting (3) drives into the unit instead of (4).

Always desktops, rarely laptops

[Proc6]

With over half the PC's sold being laptops and nearly all laptops' RAM and HD being just as user replace/upgradable as any desktop, reviewers should really give the laptop world some love.

Enough with speed. More capacity and reliability.

[sdo1]

I'm getting really tired of hyper-speed, super cheap drives that fail after a year. I've got 100's of gigs of media (ripped DVDs, ripped CDs, etc. etc.) that DO NOT need incredible latency or access speed numbers. Give me 5400 RPM drives (or slower!) that run cool and reliably! I'd imagine that most users are in the same boat. If you need a 200G drive, it's not because you need 200G for applications and games. It's because of media.

Capacity, yes. Increase that. Reliability, yes. Improve that. But hard drive speed is a grossly overrated and mostly unneeded attribute.

-S

Deathstar - top performer, yeah right.

[sl4shd0rk]

Just cuz you can get it up fast, doesn't mean it's not going to crash in 5 minutes. I've seen so many of these hitachi deskstar drives bite shit that I'll never buy another one. Even if it's free.

Re:Deathstar - top performer, yeah right.

[superpulpsicle]

I truely believe the lifespan of the drives shortening has to do with powering up and down the PCs repeatedly. Ever since I began leaving my PC running 24x7, the disks have been lasting so much longer.

If that theory is not true, then I owe it to the IBM deskstar 80GBs and 60GBs. I have multiple of these drives and they have lasted years.

Re:Deathstar - top performer, yeah right.

[Alderin1]

The last Hitachi drive I bought not only was DOA, but killed the brand new motherboard it came with, and an older replacement motherboard I had, before I narrowed it to the drive.

So, yeah, not even if it's free... in fact, not even if they paid me.

Reliability

[Anonymous Coward]

According to filtered and analyzed data collected from participating StorageReview.com readers, the Seagate Barracuda 7200.9 is more reliable than (insufficient number of samples to determine)% of the other drives in the survey that meet a certain minimum floor of participation.

Nice.

Re:Reliability

[darkwhite]

I'd mod you up, if I hadn't posted :)

Re:Enough with speed. More capacity and reliabilit

[macslut]

I'm right there with you. I have a handful of 250GB drives and a couple of 400GB drives. I would pay *more* for my drives to be 5400...even that is overkill. I want them big, quiet, cool and reliable. Speed is simply a non-issue *for me*.

WD has been great to me.

[YesIAmAScript]

I've bought WDs almost exclusively for the last 3 years or so. No problems, with drive in all kinds of devices, including 4 of my computers.

I've never been happy with Maxtor, ever since I first had problems with their 330MB and 660MB 5.25" drives back in 1988 ('89?). Their first 3.5" drive (200MB, I believe) was a loser too.

Re:WD has been great to me.

[slaker]

In 2003 I RMA'd 14 WD drives, which is hilarious, 'cause I only actually owned nine WD drives.
They're OK for you, great. Me? I like my data.

RAID improves latency, in a way

[RedLaggedTeut]

Well ok, RAID does not improve latency when you are reading that special random single byte and then send the drives to sleep again, but it does improve the latency required to fetch a big chunk of data of fixed size. You also would have the option to choose smaller drives with faster head positioning. Of course in the real world, better drive parameter often have to be purchased hand in hand with a bigger drive.

I'm off Maxtor-Bad HD's and PSUs.

[Anonymous Coward]

I've simply have had bad luck with HDs. I started with WD's a long time ago and I got six years out of them. My first mistake was with IBM Deathstars. Then I moved to Maxtor. Not only did I get stiffed on the rebate. They only lasted a litte over a year. Now I'm on a brand new (today) two SATA seagates. I also replaced the power supply because it kept dying everytime I started the machine. Now I'm using an Antec sp-500. I suspect some of my problems came from less than ideal ventilation, and poor or inadequate power supplies.

RAID claims and reality

[Jamesday]

How does RAID improve the latency or throughput of a database server transaction log file which is being flushed to disk at the end of every transaction?

Re:RAID claims and reality

[Alien Being]

A log disk will probably be dedicated to that single purpose, so as far as that particular storage system is concerned, it's just single-tasking.

There's still a potential benefit though. Some transactions will fall at the end of a cylinder, requiring a second seek. Others will be large enough that they will also require extra seeks. If the log were being written to RAID stripes, a two-seek transaction could be done with half the latency.

But like I said, the benefits vary with data access patterns. A log file is just a pathologically bad case.

in september i built

[circletimessquare]

a dual core box, with two of the 250G ata hitachis in raid0

which gives me a 465G volume at 112,844 KB/s, according to nero

i'm very happy, i edit hdv video... the raid0 volume serves as a "scratch pad" for saving video and for editting it: when you need the space

when done, the work gets saved to more reliable, slower media

it's really not that loud (in a thermaltake box with 7 fans, which is well built acoustically)

Moving day at Maxtor.

[Anonymous Coward]

Two things.

One I was shopping for new HD's (damn Maxtor), and noticed that the Hitachi's had a three year warrenty.

Two now I need to figure out a way to move W2K from a 120 to a 200 without buying Ghost.

I need to do the same with my other 120 which has Suse 9.1 on it, to a 250.

Re:Moving day at Maxtor.

[v1]

maxtor reliabiility seems to have gone up notably since they bought out quantum. all new maxtor drives are of the quantum fireball design.

Two now I need to figure out a way to move W2K from a 120 to a 200 without buying Ghost.

Not sure if it'll work, but if you format the target drive, and boot say, OS X 10.4, you should be able to

sudo ditto -rsrc /Volumes/source /Volumes/target

I think that'll work on NTFS with 10.4

Re:Moving day at Maxtor.

[Anonymous Coward]

Two now I need to figure out a way to move W2K from a 120 to a 200 without buying Ghost.

Not sure if it'll work, but if you format the target drive, and boot say, OS X 10.4, you should be able to

sudo ditto -rsrc /Volumes/source /Volumes/target

I don't know if this will work for W2K but here's how I back up my Win98 drives:

1. partition the target drive and make the main partition active.

2. format the target drive

3. set the drive jumper to select Slave mode

4. turn off power and install the drive in parallel with the master

5. turn power on and set the bios to recognize both drives.

6. add a line in the autoexec.bat to activate smartdrv with a 2 meg cache.

7. boot into windows.

8. go to MS-DOS mode.

9. use xcopy32 to copy everything to the slave drive.

10. add the system files.

The reason this works is Microsoft allows you to copy all the windows files to another drive for backup purposes. But if the second drive has system tracks, MS won't copy the needed system files.

If you have a freshly-formatted drive, MS is very happy to copy everything as fast as possible. Smartdrv speeds the process considerably. It takes about 10 minutes to format, copy, and add the system tracks in a 2 gig partition on a 466MHz machine.

I tried to include my batch file that does the work, but slashdot's lameness filter wouldn't accept the junk characters. Here's the xcopy command that does the copying:

xcopy32.exe c:\*.* d:\*.* /c /e /h /k /v

When that is done, just do sys d:

Hope this helps.

Mike Monett

Re:Moving day at Maxtor.

[Anonymous Coward]

9. use xcopy32 to copy everything to the slave drive.

xcopy does not properly clone a windows installation. See http://home.att.net/~navasgrp/tech/clone_copy.htm# Cloning for why and recommendations of tools that do clone reliably.

re: WD and Maxtor, etc.

[King_TJ]

Well, I'm somewhat inclined to agree with you. Though I never expected a lot from Samsung, my parents have one of their older IDE drives in their machine to this day that still performs just fine - at least 6 or 7 years now. Only problem now is, it just doesn't have enough capacity for modern OS's and software.

Meanwhile, I was always a traditional supporter of both WD and Maxtor (because frankly, I used a lot of Seagates and always had crashes/failures with 'em), but the warranty situation today is CRAP!

1 year only on any new retail-boxed Maxtor drive?? 1 year only on any WD drive that's not labeled a "special edition"? What's so "special" about a decent quality of build??

My 250GB SATA Maxtor DiamondMax drive just died on me, only 3 months past the 1 year warranty - and I'm now counting the days before the second one of mine does the same - since I bought both from the same place, at the same time - and both have the same manufacture date stamped on them.

I actually went back to buying Seagate for my newer SATA drives, because they're backing them with a 5 year warranty. It seems like the rules of "who is good, and who is junk" have changed around in recent years - so we'll see. If not, then at least Seagate will be swapping my drives out for free for me every so often over the next 4 or 5 years.

Re:Enough with speed. More capacity and reliabilit

[petermgreen]

until recently i'd have said a maxtor maxline but it seems that range has moved to 7200 rpm now

Re: WD and Maxtor, etc.

[Cecil]

Same here, except I switched back to Seagate about three years ago now. Couldn't have been happier. I think Seagate has really taken up the torch for reliability lately. They used to be bargain basement except for their SCSI lines, but nowadays I think they're the most solid drive you can get. I'm staying far away from Maxtor and WD, both of whom I've had recent and ongoing bad experiences with.

Is the comparison with NCQ fair or even useful?

[Anonymous Coward]

I see lots of benchmarking with and without NCQ enabled, and it appears that people are completely missing the point. If you have the option (and it works properly), you should never run a disk without NCQ! Benchmarking them together is a real disservice to all who don't understand the purpose of tagged command queueing.

NCQ allows the OS to know what has been committed to disk, which is very important from a reliability perspective. File systems do not function properly without this assurance, and can be seriously damaged on power failure.

To be fair, comparisons with NCQ should be made when write caching is turned off. Only in this case do you get the same level of reliability. Of course, ATA will be completely slaughtered, but it is a fair comparison. This abysmal performance led to the use of write caching; increasing performance at the expense of reliability. Now that it is possible to restore the reliability with NCQ, making a comparison without clarifying this point is not at all helpful.

The thing I would like to know is which disks actually implement NCQ properly, and which still lie to the system? Since drive manufacturers have been "cheating" for years on their IDE drives, has the situation truly been fixed? Spindle speed aside, it should now be possible to achieve the performance AND reliability that SCSI devices have offered for years. Unfortunately reviewers never seem to address this aspect.

Re:Is the comparison with NCQ fair or even useful?

[cbiltcliffe]

NCQ allows the OS to know what has been committed to disk, which is very important from a reliability perspective. File systems do not function properly without this assurance, and can be seriously damaged on power failure.

Whoever told you this is wrong. NCQ is nothing to do with the OS. It's a technology which enables the drive to reorder commands that are coming from the controller, in order to improve speed by, for example, taking:

Write to track 3
Write to track 94765
Write to track 4
Write to track 94733

and turning it into:

Write to track 3
Write to track 4
Write to track 94733
Write to track 94765

Less head seeks equals quicker response. The "knowing what has been committed to disk" thing is filesystem journalling. That's built into the filesystem itself, if it's supported. NTFS, Reiser, and ext3 support journalling, FAT and it's variations, ext2, and others do not.

http://en.wikipedia.org/wiki/Native_command_queuei ng
http://en.wikipedia.org/wiki/Journaling_file_syste m

Re:Is the comparison with NCQ fair or even useful?

[bani]

you are confusing command queuing with journaling filesystems.

Move your Windows to the new drive

[Nintendork]

Remove your old drive and do a fresh install of Windows on the new drive. Bring it up to the same service pack level of your old install. Boot off your old drive with the new one connected as slave. Do a backup of C and system state to a backup file on the new drive (E or whatever Windows automatically assigns to it). Boot from the new drive as master witht he old drive disconnected and do a restore of the backup file to original location and choose to always replace files. Use DBAN to wipe the old drive and sell on Ebay to recover some of the cost on the new drive. Done.

Marketing gone crazy

[Andy Gardner]

Why would anyone want a fluffy hard drive??

...........What, why are you looking at me like that

Ooooooh, flurry

reliability: the cost of offshoring?

[nido]

I care about reliability (gone down hill since 2000)

2000, eh? That's about the end of the dot-com boom. So it fits, I guess...

Since the technology bubble burst, "there's been a scramble to move from the high-cost sites [North America/Western Europe -me] to the low-cost sites in China," said Flint Pulskamp, an analyst with the market research firm IDC. "In the late 1990s, Solectron used to make motherboards for Intel right here in Milpitas, in the heart of Silicon Valley. That's unthinkable now."

-http://www.crmbuyer.com/story/44242.html

high-cost site: quality
low-cost site: quantity

anyone got a link for when various manufacturers started moving offshore? Like, Maxtor/WD/Seagate/etc.? Many computing technologies originally came from American companies, but, with cutthroat competition being what it is, as soon as one company in a category moves to china, the rest probably aren't going to be too far behind...

Re:Enough with speed. More capacity and reliabilit

[toddestan]

I agree. I bought some Samsung 160GB 5400RPM drives back when they were being phased out. Cool, quiet, reliable - great drives.

I think it's time for the Quantum Bigfoot drives to make a comeback. With today's technology, I'm sure we could easily have a 1TB drive with 5.25" platters. I'd buy one. I wouldn't really care about speed or latency issues, as I would certainly have a fast 3.5" drive to boot the OS off of.

Move your Windows to the new SATA drive

[Anonymous Coward]

Interesting. But two problems.

1) No floppy drive, but machine does have a bootable SCSI CD-burner and a SCSI DVD reader.

2) Windows 2000 doesn't have SATA drivers (those are on the MB driver disk). This is going PATA-->SATA for both operating systems.

The MB is a MSI K8T Neo-FSR so the BIOS picks it up.

I do have a Knoppix disk around here somewere.

Re:Move your Windows to the new SATA drive

[Nintendork]

Make a custom windows 2000 boot cd that includes the controller driver. http://www.thetechguide.com/forum/index.php?showto pic=5090

Re:storagereview; one of the few smart review site

[Deliveranc3]

Agreed... These things should all have been put into a database along with all the specs of the system used YEARS ago...

Of course these review sites are often making a bunch of money off this stuff.

Sites like Rotten Tomatoes exist basically leeching off movie and music reviews you'd think someone would do this for computer hardware...

I don't think it's ethical and I think stuff is going to slip through the cracks if they move to such a model but it will go a long way towards demonstrating which Hardware sites are on the level and which aren't as well as offering true comparisons accross generations which is one of the hardest things for hardware sites to show.

I like my data too

[YesIAmAScript]

Ah, can never get enough of that. If a person disagrees with me, they're stupid. In this case apparently I like losing data.

But no actually, it's been that I have had great experiences with WD over the past 5 years, basically since the Caviar series came around.

I've never had a Samsung drive. Nothing against them, just happen to have never had one.

Re:Enough with speed. More capacity and reliabilit

[evilviper]

I'm getting really tired of hyper-speed, super cheap drives that fail after a year.

They only fail after a year if they are in a rather enclosed space, with no airflow. Get a case with accomodations for a 80/92/120mm fan in-front of the hard drive areas, and you'll have far fewer problems.

Additionally, you should spend a bit more and buy from Seagate. They are typically lower power than others, and they have 5-year warranties on everything, which bodes well for their reliability.

I'd imagine that most users are in the same boat.

You'd imagine wrong. How long do you think it would take you to completely back-up your slow 5400 RPM hard drive? Are you even doing any backups?

But hard drive speed is a grossly overrated and mostly unneeded attribute.

Not true at all. Many people are spending hundreds of dollars getting faster CPUs and RAM, when they are actually HDD I/O-bound. HDD is the biggest bottleneck for most uses of computers. I know my 7200RPM hard drive is much slower than I'd like, for editing and re-compressing video. It would be much less painful if I didn't have to wait so long to write a 4GB video to disk. And 4GBs is nothing. DVDs hold 9GBs, and people are burning them left and right.

Capture a TV show, edit out the commercials, and burn it to DVD... If you're fairly fast at the editing, you'll find waiting for the hard drive to catch-up to be a big annoyance, and a big waste of your time. 9GBs of RAM isn't exactly a practical alternative.

Be glad that you don't need speed, but don't tell the rest of the world that THEY don't.

Re:storagereview; one of the few smart review site

[evilviper]

Trying to install Linux on a laptop with nocdrom or Ethernet but DLINK usb wi-fi. I NEED HELP!

If you want help, you need to explain further... Journal entries are good for that kind of thing.

NAS reccomendation

[Alderin1]

If you have some older hardware about, check this out:
NASLite

A customized Linux bootable floppy specifically designed to turn an older machine into NAS (in 5 minutes they claim), even bypassing the BIOS to support larger hard drives on motherboards that couldn't otherwise see them. Unfortunately, it does not support RAID.

Sadly, shifting in storage destroyed the motherboard I was planning to use for my own NAS system, so I can't currently give any personal experience comments.

Re:NAS reccomendation

[quis]

I've been using NASLite in a PII 300 box with 64Mb of ram and one 160Gb HDD and it works a treat. The forums over the the website the parent mentionned are also really helpful and friendly.

Collective Noun

[el_womble]

Is 'flurry' the correct collective noun? Wouldn't a spindle or harddisk review be more apropriate?

RAID 0 absolutely doensn't reduce latency...

[YesIAmAScript]

RAID 0 has no redundancy. You have to read each sector from the one point it is on each disk. That means you must seek each disk to the place and then read.

You can do all the seeks in parallel, but without spinding locking, you're gonna need to wait longer (on average) for the sectors you need to rotate under the heads, because you're going to need every read to rotate under the heads, not just half of them.

This is very similar for any RAID other than 1. I mean, with RAID 4 or 5 you still need to bring virtually all the sectors under the head (80% in a system with one parity drive per 4 drives).

I guess in theory with RAID 1 you only have to wait for the first sector of each pair to come under the head. If controllers do that, good on them. I can't imagine they all do. I mean, the overhead in cancelling half your operations once the first of each is completed would be non-trivial.

Re:RAID 0 absolutely doensn't reduce latency...

[Alien Being]

"You can do all the seeks in parallel, but without spinding locking, you're gonna need to wait longer (on average) for the sectors you need to rotate under the heads, because you're going to need every read to rotate under the heads, not just half of them."

Almost right.

"you're going to need every read to rotate under the heads"

But they're spinning in parallel. The average time it takes for a sector on one disk to reach the head is 1/2 of a revolution. The average time it takes for two sectors on separate disks to reach their heads is still 1/2 of a revolution. The point I'm making is that in that 1/2 rev, you have performed two operations instead of one.

"I guess in theory with RAID 1 you only have to wait for the first sector of each pair to come under the head. If controllers do that, good on them."

Any controller worth its salt will do this. Implementations aside, it's a natural property of RAID that it's *possible*. That's why I hate hearing people using the bogus rule-of-thumb "RAID doesn't improve latency.

I should have posted this paragraph in my first post: from the wiki page on RAID(level 1).

When reading both disks can be accessed independently. Like RAID 0 the average seek time is reduced by half when randomly reading but because each disk has the exact same data the requested sectors can always be split evenly between the disks and the seek time remains low. The transfer rate would also be doubled. For three disks the seek time would be a third and the transfer rate would be tripled.


It goes on to say:
The only limit is how many disks can be connected to the controller and its maximum transfer speed. Most IDE RAID 1 cards use a broken implementation and only read from one disk so their read performance is that of a single disk.

So maybe the rule of thumb should be "broken RAID does not improve latency".

inexpensive is relative...

[YesIAmAScript]

You have to understand what RAID meant when it was created.

The alternative to RAID was special rack-mount drives, like a Control Data Corporation SMD Sabre drive. These would cost $10K or something because they were special high-capacity units made to a higher reliability spec. Why not use a group of cheaper SCSI or ESDI drives meant for a workstation or even a PC? Since these were only $1K or something, you could put a few of them together in the right combinations to get the amount of storage with the right level of reliability you needed.

This is what Patterson and Katz were thinking, and they were right. Really, the idea of using mass production subassemblies to create systems capable of replacing "big iron" spread out and permeated the entire business. All but themost exotic high-performance systems use the same DIMMs, disks and even processors (well, families) as commodity systems now.

So, even a 15K drive still isn't outside the world of RAID. If the drive had its custom form factor, platter size and possibly its own attachment interface, then it would be non-RAID.

Re:inexpensive is relative...

[BestNicksRTaken]

Exactly.

That's why it was originally called "Redundant Array of *Inexpensive* Drives", not Independent.

The whole point was to make a large "virtual" drive for a lot less than a buying a single huge drive, performance wasn't the primary concern.

New western digital Raptor - think not

[owlstead]

Ok, so I started reading (well, I tried to between the various popup windows) and immediately found the one about a new Raptor drive. So I read into the article and it seems that they are talking about the RE2.

Differences: RE (raid edition) & RE2 specialize in "limiting the drive's error recovery time", and are specialized for raid configurations and reliability. Raptor on the other hand is speciallized in latency and reliability.

And that's when I stopped reading (and clicking away the - cannot find add server popups). Maybe it is a fast drive, but it has nothing to do with Raptor.

My lineup

[BestNicksRTaken]

I usually go by:

Seagate - most reliable, quietest, longest (5 year) warranty;

Maxtor - cheapest, they moved from a 3 year warranty to 1 year as most drives would die after 2 years;

WD - not always compatible (weird CHS values), fastest, most expensive;

IBM - they gained the "Deathstar" nickname for good reason!

Not sure if the IBM position has changed now they bought Hitachi, I doubt it as Hitachi used to concentrate on 1.8-2.5" drives, so size not reliability/performance counts.

Re:My lineup

[lazy321]

Hitachi bought IBM disk business also Hitachi first presented 400G disk http://slashdot.org/hardware/04/03/12/0440237.shtm l

Maxtor is after the lucrative prOn market

[spudnic]

Maxtor MaXLine III (300 GB)

High-fapacity competing enterprise unit (100 GB/platter)

Re:Performance vs Noise - frequency is an issue

[Anonymous Coward]

It is not just the loudness of the noise. Exposure to high frequency sounds like hard drives can cause nerve damage. I know a sysadmin who went through this a few years ago. He started hearing phantom noises outside of the server room.

NEW Forum ! On All Things Holographic

[fedrive]

Holographic Storage, monitors, lenses, Blu-Ray, communication, etc.

http://www.holoforum.com/index.php

Re:Enough with speed. More capacity and reliabilit

[jridley]

I think it's time for the Quantum Bigfoot drives to make a comeback. With today's technology, I'm sure we could easily have a 1TB drive with 5.25" platters. I'd buy one. I wouldn't really care about speed or latency issues, as I would certainly have a fast 3.5" drive to boot the OS off of.

Absolutely!

I want one fast drive on my system, the one with the swap space, the OS and the apps installed. Then I need about 1TB of space which can be very slow by today's standards. I just want it to be very reliable and big.

Reliability

[Anonymous Coward]

Disclaimer: I have previously purchased hard drives for a living. This is based directly on personal experience. Your personal experience, for reasons I will explain below, may well differ.

Bad drives, from all manufacturers, come in batches. A batch will either be Dead on Arrival (DoA), Dead About A Week After Arrival (faulty), Dead After About A Month (flaky), or absolutely fine (good).

The batches are highly localised to supplier/region/manufacturer, and one drive in a batch is fairly representative of the batch; if one drive is bad, view the whole crate with suspicion. If you get stuff too far down the supply chain to get them batch-by-batch, then you'll just have to cross your fingers.

For example: Joe might have never had any problems with Maxtor, and will swear by them, but every Western Digital he's ever had dies in the first week. Until he switches supplier, or moves. Now, suddenly, he gets three consecutive bad batches of Maxtors, but the Western Digitals are fine. Paul, on the other hand, in another country, swears by Seagate, so Joe tries Seagate, and the whole crate of drives arrives DoA even though they're normally fine.

Point is this: Bad drives come in groups, no manufacturer is immune to them, and no manufacturer is really better than any other overall (but supplies in your country, or to your supplier, may tell a very different story than the overall supply).

Fast drives are slightly less reliable than slow drives, but fast drives are also manufactured to tighter tolerances. It roughly balances out, but heat is more of an issue with fast drives. Moreso with 15K ones. The Raptors seem to be an excellent sweet spot for a desktop drive.

Hot drives die quicker. Drives with more platters run hotter. Drives with no airflow past them run hotter again. Drives with no airflow past them at the top of a stack with no airflow past them will die no matter what. Drives which are faulty, will die prematurely, regardless of all the above. With a burn-in, you might be able to tell DoA and faulty drives apart from the others, but it's almost impossible to determine which ones are flaky and which ones are good. Your best bet is to look at the SMART data; but on at least half of flaky drives, they won't show the tiniest bit of problem until right before they die catastrophically.

Excepting extreme outliers like the Deathstars, wihch were doomed from the very beginning (and those were, I'm afraid, only the GXPs; modern Hitachi Deskstars like the popular 7K400 and the high-capacity 7K500 are actually surprisingly good in the reliability stakes, given the 7K500 does run hot so needs airflow), that's pretty much it.

Don't get cheap NAS units (RAID arrays in a box) or RAID controllers. Either 3ware or Adaptec, never Promise, Highpoint or VIA. Cheap RAID controllers die more often than RAID arrays. Software RAID is far more reliable than cheap hardware RAID.

RMA varies between manufacturers... and regions, so I can't give you useful advice here, except that RMA policy is more important than warranty in many situations. Don't forget that some suppliers offer better RMA policies direct than the manufacturers; but some, very much the other way around.

Manufacturers' general traits (your experiences may differ): Maxtors click when seeking and run hot, and click more and whine when they are about to die. WDs don't click when seeking and don't whine, and also run fairly cool. Hitachis are fast at random seeking for their class, and come in larger capacities (7K500). Seagates are slower, and a bit clicky and whiny, but some people have better reliability with them (as above, some people have worse, it's luck). Samsung Spinpoints are slow, but the quietest.

And remember: All drives die. It's just a matter of treating them right, and beyond that, just luck, and time.

Re:Enough with speed. More capacity and reliabilit

[WuphonsReach]

Capture a TV show, edit out the commercials, and burn it to DVD... If you're fairly fast at the editing, you'll find waiting for the hard drive to catch-up to be a big annoyance, and a big waste of your time. 9GBs of RAM isn't exactly a practical alternative.

That's basically what I do on my one box. Capture straight to MPEG2, then cut the commercials in TMPGEnc DVD Author prior to putting it on a DVD. My source material is around 11.5GB for 3.5 hours (I record 7:45pm to 11:15pm every day, then chop to what I keep).

Right now, that box has 7200rpm drives and 1GB of RAM. Cutting is laggy, until system caches enough of the file into RAM. Moving from point to point on a fresh file takes a second or three for it to read the file and thumbnail, then it gets pretty snappy.

Even with re-encoding the video to AC3, I'm still disk-bound. Even when my source MPEG files are on one 7200rpm drive and my target drive for the DVD build is on a 2nd 7200rpm drive. The dual-Opteron 246 CPU utilization is typically ~10-20% and it takes me about 16 minutes to create the DVD with a pair of 43 minute episodes.

I'm debating the $250 for add another 2 sticks of RAM to the system (taking it up to 3GB). That would give me a much larger disk cache and is a cheap upgrade for this unit. Plus going from 1GB to 3GB gives me more headroom for other software. Moving to 10k rpm SATAs might be an option as well, but a more expensive fix.

And under 100 pounds even in the UK!

[rklrkl]

I just ordered the Maxtor Maxline III 300GB from Ebuyer - 91 pounds with a 5-year warranty (I see a load of people bitching about Maxtor's 1-year warranties, but this baby has a full 5 years on it). Yes, it's got a 16MB cache (Ebuyer's one-liner for the product is wrong - they get right lower down the page). Don't forget to buy a SATA data cable of course.

And, no, it'll be used to rip my entire CD collection to MP3's before you ask...