The Blog of Matt (be afraid)

For several years now SSDs have been the up-and-coming storage technology. In the past couple of years, their development has reached a fever-pitch level as the manufacturers of solid state memory continue to find new ways to push more transistors into smaller areas, and the respective controllers continue to squeeze more performance per watt and refine logic on the wear-leveling algorithms. All this is well and good, but I really sincerely believe that pushing these into the hands of every-day users is probably one of the worst decisions from a technology standpoint the industry has made thus far. This is not to say it’s a bad BUSINESS decision, as the flash manufacturers stand to make a lot of money if they can supplant the ubiquitous hard drive, but that doesn’t make it the right decision.

First, here is brief history for the readers who might have stumbled upon my 2^nd blog entry and are saying “Just what is this cat talking about?” Solid state disks (SSDs for those in the industry) are, for lack of a better description, an extremely close relative to the storage cards (SD, CompactFlash, etc.) you use in your digital camera or cell phone.  The differences lie in the minutia of manufacturing, and that most SSDs utilize onboard controllers, while most memory card technologies have their controllers on the device. Onboard controller are useful because they reduce the delay induced by having to make the calls to the NAND flash that in memory cards happens over most times a friction-based connection. For the sticklers out there, yes there are solid state drives based upon volatile DRAM which are quite different in technology, and will be addresses in a later article. NAND-based SSDs are going to be the topic of conversation, since they are becoming more wide-spread as the days roll on. For more information, Wikipedia has excellent resources on solid state drives.

I have always been an outlier when it comes to recommending upgrades. The first two things that come out of the typical IT person’s mouth is usually “Processor” followed by “RAM,” with the two flip-flopping positions depending on who you talk to. I have always recommended hard drives as being the key place of interest to be spending money to increase speed. I once read a statistic that if we were to scale the amount of time a processor spent waiting for the disk to retrieve data to a number we would grasp, it would equate to around 100,000 years.“But why, Matt? Processors and RAM are so easy to upgrade and the difference is usually noticeable immediately.” Ah, that might be true, but with the advent of cheap huge storage devices, like hard drives, external NAS, and big RAID arrays, more and more programmers are beginning to advocate that utilizing the disk is not a huge issue, since the space is there to do it. Windows Vista/7 shipped on a DVD, and Red Hat takes 6 CDs or 1 DVD, just to give a few examples. So our reliance on hard drive storage is becoming greater and greater.

The largest problem with hard drives is their rotational latency. Since it’s a moving part with a few read heads, the drive has to be spun to the spot where the data is stored, and the head has to move there to read it. The more data you pack onto a hard drive platter, the more data is in a particular square centimeter, and thus the head doesn’t have to move so much to gather a lot of data. This has seriously improved the 3.5” drive market, and been a huge boon to the 2.5” drive market, as the enterprise versions of those drives are reaching speeds equivalent to the 3.5” drives. Solid state disks though far surpass the performance level of today’s spinning hard drives, as the data is available instantly. And even if the fastest 15k drives can still spool data faster once they have searched to it, the smaller files are basically instantaneous on a SSD while you are still waiting for a normal drive to find the same file and begin to transfer it.

Now, I just sat here and typed a whole page to what SSDs are basically going to be the next best thing since sliced bread (for computers anyway). Why the dismal title of it being not a good thing for the masses? If I had to phrase it up in a word, it would be “size”. SSDs are never EVER going to match up with classical hard drives when it comes to price per gigabyte (soon to be terabyte). 1.5TBs are quickly approaching the $100 dollar barrier and below. 1TB SSDs are still floating around 1,000 dollars, and are in such short supply that you will probably never be able to actually acquire one unless you are a medium to large OEM.

Yes, I used the word “never.” In fact, never EVER was the phrased I used to describe and I’ll tell you why. Hard drive manufacturers are going to continue scaling their technology at a rate as fast if not faster than SSD manufacturers because they are working with proven technology that all they have to do is tweak, whereas SSD manufacturers are plagued by manufacturing defects and controller issues. This is not to mention the huge supply chain issues they face as they try to scale the size of the drives up to compete with classic hard drives. A classic hard drive spins on a metal composite platform with a metal coating that is very weakly magnetized so that the read heads don’t adhere to the surface and can change the polarity of the individual sectors to indicate a 1 or 0. SSDs on the other hand require very specialized silicon that is in short supply already because of the stresses of other manufacturing processes that utilize it.

So what is the solution? Well, we had one that everyone just seemed to graze by: hybrid drives. These utilized a cache of memory that front-ended a spinning drive platter that was used for storage. This technology was pioneered by Samsung and has now basically been forgotten by most of the tech industry. Part of the problem was the blending of two separate markets (Samsung happens to be a part of both), and part of the problem was the lack of logic in the software world. Windows specifically had no idea how to take proper advantage of the technology, which is just a lack of proper programming on their part, but that’s not to say everyone else didn’t completely miss the boat as well. Now we have this raging feud between the flash producers versus the hard drive manufacturers, and in the end the consumer is going to lose because if the two groups had pooled their development dollars into the hybrid drive technology, we could probably have the best of both worlds at this point.

Well I think that’s enough for this week. Have a great week everyone, and enjoy the holiday for everyone who celebrates it here in the USA. Tune in next week.