Sign in to follow this  
mcg

Why are 160 GB HDDs so cheap?

Recommended Posts

To expand upon that a bit more... a 120GB drive is just a 160GB drive with 40GB that you can't use. While this does allow the manufacturer to use some platters that would fail to qualify for the higher capacity, the savings is not so much.

Share this post


Link to post
Share on other sites
To expand upon that a bit more... a 120GB drive is just a 160GB drive with 40GB that you can't use.  While this does allow the manufacturer to use some platters that would fail to qualify for the higher capacity, the savings is not so much.

I assumed the 120s were the 160s with a platter side not being used.

While this leaves a platter side that doesn't need to be manufactured, I've never considered this might mean a platter that didn't meet the quality specs for a 160...

Besides, my thought would account for why some 120s are cheaper than 160s (one platter side not being manuf), and account for why some 120s cost more than 160s (120 and 160 are identical, one platter shut off for the 120, and since they cost the same to build, but the 160 has higher demand, less "120ed" 160s are made, hence the higher price again...)

Besides, some 120s ARE three 40GB platters... Or maybe two 60s... (I know my IBM 120GB drive is not a crippled 160...)

Share this post


Link to post
Share on other sites

I would assume that there was a problem with one side of a platter, bad head?

anyway, it is more impressive looking that the cost going from 40GB to 80GB drives. One week I noticed the normal prices such that the 40GB had a higher asking price (lower volumes?) It was only a few $, but still.

Share this post


Link to post
Share on other sites

I picked up 2 WD1600JB's for $89 apiece a month apart. This was at a brick and mortar store with no rebates required (PC Club). I got to look at the drives before purchasing to ensure they featured FDB motors :D

Share this post


Link to post
Share on other sites
Besides, some 120s ARE three 40GB platters...  Or maybe two 60s...  (I know my IBM 120GB drive is not a crippled 160...)

That was then, this is now. When I bougt my 40 gb 60 GXP, it was also the one with the best price/capacity ratio (two-platter version).

When I bougt my 120 gb 180 GXP, it was again the one with the best ratio and also the two-platter version.

BTW, if there's more demand, prices are usually higher, not lower.

Share this post


Link to post
Share on other sites

Two-platter, four-head drives have offered the most capacity for the money for quite some time, possibly for even the last decade. Most of my drives are two-platter, starting with the 8.4 GB I purchased in 1999. My next purchase will probably be a 200GB, 2-platter design, and I'll wait until CompUSA or someone else has a special for maybe $79.99 or thereabouts.

As for the 40GB one-head drives, I personally don't see why anyone would buy them. You save maybe $10 over 80GB drives, but get half the capacity. Ditto for drives with "destroked" platters. You're already paying for the head and the platter-you may as well use all of it, and both sides. I don't see why "crippled" things like this are even made, I really don't, or why anybody would buy a 40GB drive which costs only a little less than an 80GB one. Even more so when you'll see them replaced less than a year later because they didn't have enough capacity. Where was the savings? :D

Share this post


Link to post
Share on other sites

I would say that the primary reason that drives with odd numbers of heads, or short stroked platters, make it to market, is that they allow hard drive manufacturer's to make use of media that they may have otherwise needed to toss out.

There is presently a lot of technology and development going into this field amongst drive manufactuerers. Margins are so tight that they are trying to minimize waste as much as possible. You may have noticed that it has become difficult to determine exactly what media a drive of a given capacity may be using (Western Digital and Maxtor are the biggest culprits here). This is because manufacturers are not throwing out platters that fail to make the cut for a given density, and reuse them at a lower density. Remember trying to figure out if you were getting a 66GB/platter, a 80GB/platter, or a 100GB/platter JB from Western Digital?

Samsung is now taking this to a whole new level. Samsung uses low level formats to change the data density of each surface in the disk. This means they need to translate linear addresses differently, because tracks on different surfaces won't necessarily line up, due to the per-surface formatting. This allows samsung to mix and match platters with tremendous flexibility at the cost of a little firmware complexity.

It also means it is going to become very hard to determine the low-level performance characteristics of hard drives, since they will vary from drive to drive within a family to an even greater extent than they already do.

Share this post


Link to post
Share on other sites
I would say that the primary reason that drives with odd numbers of heads, or short stroked platters, make it to market, is that they allow hard drive manufacturer's to make use of media that they may have otherwise needed to toss out.

That explanation certainly makes sense to me. My puzzlement though is why anybody would buy them. I checked on Pricewatch. The cheapest 40GB drive is $38 while the cheapest 80GB goes for $46. I can't imagine even a pauper being on such a tight budget that they wouldn't want double the capacity for a lousy $8 more. Maybe I can understand going with one platter as opposed to two, even though here the price difference isn't worthwhile either unless you're on a really tight budget (you can find plenty of two-platter 160GB drives for ~$73, or only $27 more than the cheapest one-platter drive). By the same token, unless you're pressed for space I don't much see the point of 3 or 4 platters, either, at least from a cost per GB standpoint. 2 platters is the sweet spot, and probably will be at least until platters become so dense that making two platters work together becomes much more expensive (and then one platter will be the best bargain).

Share this post


Link to post
Share on other sites

Hiy

If you imagine a company builiding maybe 40 or 50 office PCs to be used for utterly menial tasks, with none of them ever likely to use more than 6GB of the harddisk then getting the cheapest drive around, even if it's only a couple of dollars difference, makes sense. Why pay for capacity that will never, ever get used?

For the vast majority of PCs used in business today a 20GB drive is overkill - all important work is stored on network storage, so it's only the application installations that use local storage.

Cheers.

Dave

Share this post


Link to post
Share on other sites

I don't understand why they don't make 20's anymore. Then should still make small hard drives. What are offices gonna do when they start running out of 20 gig drives in their storage rooms? There's gonna be a big demand for them especially since many computers can't take a lot of hard drive space. Heck, my school still uses 350mhz p2, 128ram and 8gig hard drives.

Share this post


Link to post
Share on other sites
If you imagine a company builiding maybe 40 or 50 office PCs to be used for utterly menial tasks, with none of them ever likely to use more than 6GB of the harddisk then getting the cheapest drive around, even if it's only a couple of dollars difference, makes sense. Why pay for capacity that will never, ever get used?

For the vast majority of PCs used in business today a 20GB drive is overkill - all important work is stored on network storage, so it's only the application installations that use local storage.

While I agree, I think diskless clients are more suitable for the environment you describe. Why bother with a hard disk at all in that situation? If saving $8 dollars are worth it, then saving 5 times that should be as well.

What are offices gonna do when they start running out of 20 gig drives in their storage rooms?

Replace them with 40GB drives that cost less than they paid for the 20GB drives?

Share this post


Link to post
Share on other sites

Diskless systems must be on the way back in. The last time I used one it was run with 24 PCs on a 10Mb/s coax network. It just just runnable then, but now days a 100Mb/s switched network would give the required bandwidth in a similar enviroment.

Share this post


Link to post
Share on other sites

Back when I was in school they ran a diskless setup. By the time I left it was around 120 machines booting Win 3.1 and running all their apps (Word 2.0, Pagemaker 4) off a single OS/2 server. 10MBit Coax. You can imagine how great it was.

Dave

Share this post


Link to post
Share on other sites

I always thought that two users on coax was one user too many. Of course, remembering the crap that we used before ethernet (latticenet anyone?) coax was a "joy" to work with. Cheaper than arcnet too :-)

Share this post


Link to post
Share on other sites
Samsung is now taking this to a whole new level.  Samsung uses low level formats to change the data density of each surface in the disk.  This means they need to translate linear addresses differently, because tracks on different surfaces won't necessarily line up, due to the per-surface formatting.  This allows samsung to mix and match platters with tremendous flexibility at the cost of a little firmware complexity. 

That's a little scary to hear. That means that head-switch times within the same cylinder will be non-neglible. Then again, who buys Samsung drives for performance reasons?

Share this post


Link to post
Share on other sites
My puzzlement though is why anybody would buy them.  I checked on Pricewatch.  The cheapest 40GB drive is $38 while the cheapest 80GB goes for $46.  I can't imagine even a pauper being on such a tight budget that they wouldn't want double the capacity for a lousy $8 more.

I can't either, but don't forget - there may still be a demand for those lower capacities, because of reasons other than price/performance rations, such as BIOS bugs and other platform-specific limitations that would prevent the utilization of a higher-capacity IDE drive. It is for the same reason that items such as PIII 1Ghz "coppermine" CPUs still sell for prices rivalling faster PIV and Athlon chips, even though their price/performance ratios are much lower.

Share this post


Link to post
Share on other sites
That's a little scary to hear. That means that head-switch times within the same cylinder will be non-neglible. Then again, who buys Samsung drives for performance reasons?

As per information outlined by Freeborn a little while back:

cylinders are still a valid concept to envision the physical layout of data tracks. Drives just don't write or read an entire cylinder at once now....you would have to go back to stepper motor drives to find a drive that meets it as a geometric concept. It's a nice abstraction though that makes visualization easier.

Share this post


Link to post
Share on other sites
cylinders are still a valid concept to envision the physical layout of data tracks. Drives just don't write or read an entire cylinder at once now....you would have to go back to stepper motor drives to find a drive that meets it as a geometric concept. It's a nice abstraction though that makes visualization easier.

As far as I know, drives never did write an entire cylinder at once anyways. That would require being able to read/write from multiple heads simultanously in the device, which would require multiple parallel read/write channels, etc., to say nothing about the servo problem, since all of the heads are linked physically, but their alignment over each platter surface may vary by microscopic (but non-trivial) amounts.

I was always under the impression that (at least ever since zone-bit recording), the model was a relative performance-based model, that was used and embedded into the performance-oriented drive-control and filesystem/storage-layer software of various OSes, based on the "guarantees" of that model, one of which that head-switch time was negligible compared to cylinder-seek time. I guess that is no longer going to be the case here. It's going to break a lot of performance models for hard drives.

(I think PCGuide has a link to the model somewhere, or perhaps this site has it too? One of the other features of the performance model was that lower LBAs were faster than higher LBAs, although technically, that doesn't have to be the case, if one completely abandons the theoretical performance model that I'm talking about.)

Share this post


Link to post
Share on other sites
For the vast majority of PCs used in business today a 20GB drive is overkill - all important work is stored on network storage, so it's only the application installations that use local storage.

While I agree, I think diskless clients are more suitable for the environment you describe. Why bother with a hard disk at all in that situation? If saving $8 dollars are worth it, then saving 5 times that should be as well.

I'll also add that diskless clients are good for another reason-you can control exactly which programs the employee can run. This means you no longer have to deal with employees playing Solitaire on company time, or surfing the Internet (unless their job requires it). A completely diskless workstation has another big advantage-the person can't walk off with company info.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Sign in to follow this