Why aren't small drives cheaper?

Why aren't small drives cheaper?

Summary: Smaller Is Better, So Why Isn't It Cheaper?I've been wondering when the 2.

TOPICS: Hardware

Smaller Is Better, So Why Isn't It Cheaper? I've been wondering when the 2.5" form factor will replace the 3.5" drive. After all, the 100,000,000x increase in areal density has seen the extinction of 24", 14", 9", 8" and 5.25" form factors. Areal density keeps increasing, so why haven't we yet migrated to 2.5" small form factor (SFF) drives?

Related to that I wonder why SFF drives are so much more expensive than 3.5" drives. Check it out yourself: a 160 GB 3.5" costs about 2/3 that of a 160 GB 2.5". Here are some answers.

Why SFF drives should be cheaper than 3.5"s

  • Lower material costs. SFF drives weigh a fraction of larger drives. Doesn't metal cost something?
  • Smaller, slower motors and slower, cheaper electronics. Bit rates off the head are significantly lower at 5400 RPM, so the signal processing chips should be cheaper.
  • Shipping and distribution costs lower. A lot more SFF drives fit in a 40 ft. container, the shipping cost per drive should be noticeably lower

Not everything is cheaper just because it is smaller.

  • Read/write heads are the same, per platter, but you need more for a given capacity. R/W heads are a major cost in disk drives, which is why moving from 4 platters to 3, or from 3 to 2 is a big deal for vendors and prices. SFF disks are smaller, so at a given capacity SFF drives require more heads. Today, an 80 GB SFF drive costs the same as a 160 GB 3.5".
  • Learning curves. As production volume rises, costs drop. 3.5" drive volume is much higher than SFF. The move to notebooks is helping, but 3.5" drives go into lots of other applications, such as arrays, that SFF drives haven't penetrated.
  • SFF margins are higher. I want drive vendors to be profitable and they are making good money from SFF drives.

Anything else? I sure hope the vendors aren't getting together in some back room somewhere and divvying up the market. That would be wrong, illegal and very, very stupid.

I would appreciate it if a reader from the industry would give us their take on this in a comment or write me directly. Use an alias and bare all - I'll never tell who sent me the email. Contact info at StorageMojo.

The Storage Bits take In The Innovator's Dilemma, a must-read book for every storage aficionado, Christensen makes the case that form factor changes have been the prime mover behind the rise and demise of most disk drive companies. Now that we are down to so few vendors I doubt that will happen again. Most vendors make all sizes, unless, like Toshiba, the specialize in small form factors.

Consumers are driving disk costs today. As long as we keep choosing capacity over power and size SFF drives won't take off beyond notebooks. But every other form factor transition had the same problem. It will be interesting to see how this plays out.

Comments welcome, as always./

Topic: Hardware

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • Not rocket science

    There is no difference between "big" and "small" in terms of physical materials and labor required to make it. At say $50 for a cheap small drive, the cost is probably $25 with a $25 markup. That's the base.

    For higher capacity drives, the materials are the same, but the ROI required to make the R&D pay for itself means that the drive needs to cost more -- not that the cheapest drive needs to cost less.
  • Because you are paying for capacity.

    What you are paying for is not the physical materials (though they do have a cost) but rather for the capacity. A 160GB disk costs roughly the same be it a 3.5" or 2.5" physical size. The smaller ones are more typical for laptops while the larger ones are more typical for (legacy) desktop machines and servers (depending on interface and needs).
    • HIgher data density?

      A 160Gb small drive needs higher data density than a 160Gb "big" drive.

      This probably means more platters inside it or something.

      Try comparing the price of two drives with similar data density.
  • Doesn't it cost money to increase density?

    Doesn't it cost money to increase density? If you're putting the same amount of data onto a smaller surface, then you're probably talking smaller heads and higher technologies for the same amount of data - as well as the possibility of adding more platters, as you have already observed.

    I'm guessing that 3.5" is probably close to the optimum tradeoff between benefits and costs. Try and get a drive too small, then the costs just might overwhelm the benefits of a smaller form factor.
  • You are paying for a few things.

    You are paying for the newest technology, which is novelty. It is a must have in some markets. Early adopters never get a good deal. Wait awhile and the prices will come down. Let others do it for you.

    You are also paying for limited numbers which are covering the cost of engineering and tooling out in the shop. Stuff like the aluminum die cast dies, different stamp dies for the metal in the read write heads and circuit boards. Once these go up in numbers, they will narrow the gap. In 3.5", a 160GB hard drive is only $10 more than a 80GB. So it isn't capacity. 5400 rpm drives are about $10 less in price then a 7200 rpm. So it isn't speed.

    By the way, 2.5" drives are not better, but different for special needs. They are slower, less capacity, not as durable but consume less electricity, are smaller and lighter. Great for a notebook but not for a high capacity cannot fail server. By the way, the 3.5" disks for that server are more expensive than the retail grade but not as much as a 2.5".

    So you pay for:

    Energy Sipper
    Fewer Numbers
    Tooling Costs

    Besides I am looking forward to durable, high speed, high capacity solid state drives. 2.5" are only a moderate improvement for certain applications.
  • 2.5" drives have failed me more than 3.5".

    I've had far more failures with 2.5" drives, and far earlier too. Many of my notebook drives have died within 2 years, which is lower than the average 4+ years I get from 3.5" drives.
  • More important- why not FASTER???

    This is off topic, but related. Yes, smaller is better, but faster is better too. Capacity keeps doubling every year or two, and form factors are getting smaller, but we're still seeing the same 10ms or so seek times we've been living with for close to 10 years. Why can't somebody start selling 80G drives that cost as much as 750G drives but have 2ms seek times? Why not put 4 sets of heads at 90 degree intervals and quadruple performance, rather than just making them bigger and bigger, in smaller and smaller size boxes?
    Steve Summers
    • actually, it is faster roughly doubling every year or two

      That's the speed of the seek. If you double the capacity of a drive and double the speed of the seek guess what the seek time remains the same because you're seeking a specific head, track, sector, byte address through twice the number of such addresses. Basically, there seems to be an issue with the perception of speed rather than the reality of speed increases.
    • Small = lame

      At the moment, small = lame, and who wants lame?

      It costs about the same to build a 2-platter drive of a certain physical form factor, irrespective of the capacity. As capacity of those platters goes up, it's like a "free lunch".

      That's why you get uneven price notches between certain sizes, e.g. in a day when HDs were 20G, 40G, 60G and 80G (using 20G a side), the artificial disabled-head 20G and 60G would be not a lot cheaper than 40G and 80G respectively, once the fat is squeezed out of the pricing.

      As to "why not make a fast lame-capacity HD the same price as a big cheap drive?"; the answer is that the biggest performance hit is moving the heads. So the best way to speed up a HD is to pack more per "cylinder" (i.e. all the tracks on a certain head position) by boosting linear density and adding platters.

      I was surprised when 3.5" drives took over from 5.25", as I expected the larger area of 5.25" drives to always out-capacity 3.5" drives and thus always be preferable. Perhaps 3.5" is more mechanically efficient at higher spin speeds etc., as the only large capacity 5.25" survivors were the cheap-and-slow BigFoot drives from Quantum.

      The same logic doesn't seem to apply to 2.5" vs. 3.5", and I think other posters may have pointed out the possible reasons for this - i.e. that 2.5" development has concentrated on low power consumption and light weight (doesn't feel like big rare-earth magnets in there, for a start) and as such is still a specialized niche thing for applications where functional suckage (poor capacity-to-cost ratios, speed) are "OK".
    • Multiple actuators/drive for faster access?

      Adding additional arms has many problems that would make adding multiple full-stroke actuators less cost effective than you might think. All of the heads would need to remain in calibration with eachother, or the signals may be read differently by different heads for the same track. This is a harder problem than you might think (harken back to the days of multi-platter disk packs on washing-machine sized drives; moving a pack from one drive to another of the same model was iffy, mostly due to head calibration issues.)

      Adding to the problems of such a drive is the scheduling of reads/writes vs. heads. You must have a queue of pending requests to optimize the motion, and quite a few other factors creep in. Very few systems use actual physical geometries anymore; the drive maps the geometry that fits in the IDE/ATA/SATA control registers to the actual geometry. Management of multiple heads/drive would be embedded into the drive. The time of combined track seek / settle / rotational delay / read operations will not change for each actuator, so the only real savings is if a head is already on the track in question.

      In the bad old days, I had an old 10" fixed head "swap" drive (it might have been 14"; I'm not sure now). It had fixed heads, one per track. Track seek was pretty much 0, rotational latency was the only factor in reading a sector (discounting the switching time between heads). This had a dedicated controller, and a rather small capacity, but (for the time) it was very fast.

      Using that as a model, you could improve access speeds (on the average) by having multiple heads per surface on a single actuator. For reasons of calibration, a given track could only be read by one head, and the actuator would seek in groups of the number of heads. This way, fewer short seeks would be needed, as track-to-track time would be 0 for tracks within a group and track-to-track times that were out of group would still take about the same time they would for the 1 head/surface drives. Since only one head would be active at a time, the electronics would not have to be multi-channel, and there should be no head-to-head interference. Combine it with large on-board track caches for reads, and you've got potential for a really screaming read device (I'm discounting off cylendar/non-cached bad block replacement). Writes can't be queued and written as safely.

      I'm not in the hard drive business, so I don't know if that idea would work at all. I am pretty sure the multiple actuator drive would be a flop -- high failure rate (more parts to fail), and little performance gain.
  • In the computer biz - smaller is rarely cheaper

    Why are 2 512MB memory sticks only slightly more than a single 1GB stick? Why is notebook memory more expensive per MB than desktop memory? Why does a DVD for your laptap cost more than one for your desktop. Why do little cell phones cost more than larger cell phones?

    Cost of materials is a small factor in the retail price of a product. Manufacturing costs and competition have more effect.

    With the price of drives today, I am not too worried about collusion. I bought a 320GB 3.5" drive for $60 2 weeks ago.
    This is a retail drive, and no silly rebates. It is amazing to think how much more storage you get for your dollar than just a few years ago. A dollar a gig used to be a good deal, now I am paying less than 18 cents.
  • Economies of scale.

    Usually it costs less per unit the larger the batch. The market for 3.5" drives is larger than the market for 2.5" drives.

    Although this may change as laptops overtake the OEM desktop market.
  • Slower signal processor?

    If the newer denser technology is only on the smaller disk, wouldn't the signal rate remain the same in spite of the lower RPM?
  • Cases

    We'll sell more 2.5" drives when we have cases that provide space for them. Very few towers have a 2.5" slot I love my IBM SFF systems. I'd like them more if I could get 2 drives inside instead of just 1.
  • I thing the real point is this!...

    even if greater capacity costs another 50 cents in reality.

    they're going to charge another 100 bucks for it.

    not fair.
  • I don't know about cheaper but . . .

    The 2.5" 15k RPM SFF SAS drives in my servers certainly aren't lame (or cheap). More servers are using the sff drives these days and that may help drive down costs as well.
  • Higher price for faster access

    In answer to the person who asked about charging more for a faster drive with maybe less capacity. Actually they already have this in the form of solid state drives - the only advantage rotating media offers is that it is cheaper per increment of storage and at least for the forseeable future will be.
  • Duh: Why doesn't Intel sell 65 nm chips for less...

    2.5 drives are always single-platter. So compare a 2.5 single platter to a 3.5 single platter. The areal density of the 3.5 is going to be half that of the 2.5. This means the 3.5 can use less advanced materials (platter and heads).

    Your conjecture makes about as much sense as asking "Why doesn't Intel sell their 65nm chips for 2/3 the cost of their 90nm ones?" You don't even have to think about that one.

    It's the same thing with the drives. The smaller stuff uses more refined technology. It always will. Every time there's a generational advance, the smaller drives are at the cusp, and the larger drives adopt what HAD been the smaller drive's technologies, roughly speaking.
    • Actually, Intel DOES sell their 65nm transistors for less

      That is the basis of Moore's Law: more transistors on a chip every 18 months or
      so. As chip prices remain constant, transistor prices fall.

      The learning curve effects are why new chips are more costly than older chips +
      the small factor of demand.

      Nor is it correct to say that 2.5" drives are single platter. Dual platter is very

      Thanks for reading.

      R Harris
  • Don't read this waste of space

    Can I have that 5 minutes of my life back please? Captain Obvious alert.