In this new post I am going to expose some facts about storage drives in a computer. That means I want to talk about (and also compare) two of the most important categories of storage drives that we make use of every day – hard disk drives versus the so-called SSD drives, seeing what they are, what each of them can do, and why not, also the pros & cons for each side in order to better understand what kind of disk space our computers use, can use or should use!
What are „Solid State” and „Hard” drives?
Since computers started to exist in the world, the problem of having storage space for activities emerged naturally. The smallest storage pieces were the old „floppies”, also known as diskettes, which nowadays are rather like museum pieces. They were featuring storage dimensions that today would seem ridiculous or incredible, like 720 kilobytes (decimal kilobytes, so 720000 bytes) or 1.44 megabytes (still decimal!), and some of those had even 2.88 MB of space. But of course, as the range of desired computer activities increased, the urge for larger space led people to new solutions.
Paradoxically, chronologically speaking, the so-called hard disk drives were invented ten years before the diskettes – in 1956 IBM produced the IBM 305 RAMAC model, a hard disk drive that was… literally hard: it weighed more than a ton – yes, it is true 🙂 – and while it was able to store just 5 megabytes (again, this is not a make-up, it was 5 MB), it needed a forklift because it was too heavy for being manually carried by people. And in 1967 IBM also launched the first floppy disk drive, that was 8 inch in size (later they got smaller and were called diskettes).
But hard disks were much more useful in matter of size and by the time their capacities increased (10 MB, 20, 40 and so on) while physical sizes decreased over the next decades, so that today we may see even 16 TB hard disk drives (sixteen thousand billion bytes, almost the square of the bytes count that the first HDD had back in 1956), and an adult person can easily hold in their palm a HDD like that. However, a HDD drive still has a certain weight that somehow justifies the „hard” term.
So how does a HDD work? Hard disks are non-volatile memory hardware pieces (like floppies were too, and unlike RAM is). By volatile we mean that when electrical power is turned off, any data disappears forever, whereas in non-volatile storage environments we keep the data in the absence of power, unless we decide to erase it manually. Its storage space consists of one or more round-formed magnetic platters (that actually contain all those terabytes of space), a mechanical system with a read/write head that is able to work with the data from those platters, and also a motor for moving the platters and the head part. Since platters are round, their movement is a spin rotation, and this is what happens while a HDD is running. There is also an input/output (I/O) controller, and a firmware part (that green or blue logical circuit from the back of the HDD, called Printed Circuit Board aka PCB) that control the activity of the RW head & platters zone, ensuring also the communication of the disk with the rest of the computer.
The round platters are made of aluminum, glass or ceramic concentric circles (aka tracks) that are on their turn divided into logical units we call „sectors”. Yes, those HDD sectors that sometimes (unfortunately) can become bad and pull an alarm signal for us. As far as the platter spin rotation is concerned, for our personal computers its speed ranges from 4200 to 7200 rotations per minute (aka RPMs), and the greater this speed, the faster the HDD works. So people who want best hard disk drives, speaking in a matter of speed, should look for those featured with 7200 RPM.
The mechanical part of a hard disk drive is fragile and it gets very probably damaged if we accidentally drop the drive, so this is a drawback for HDDs. Also, speaking of the firmware / logical circuit part, it can be damaged too, like getting fried by a short-circuit, and while older HDD from the 90s allowed this circuit part to be changed with another one from a similar HDD model, nowadays HDDs are sold with unique firmware-hardware combinations, meaning that we cannot replace the damaged circuit from the back of the disk with a similar one. If something bad happens there, we need the specialized recovery data centers or simply to buy another HDD unless we can replace the entire HDD for free via warranty.
But technology continued to develop over the years, and newer & faster pieces appeared on the market. In 1991 SunDisk manufactured a flash memory-based drive, that was 20 MB in size, and had a PCMCIA (Personal Computer Memory Card International Association) configuration. It costed around $1000 US at that time, but of course it did not have any similarly huge physical dimension if we compare it to the first HDD from September 1956.
But first of all, what is flash memory? It is a kind of non-volatile memory (as we can see in the definition for HDD), used for data transfers between a personal computer and digital devices, and it has the ability of being electronically reprogrammed and erased, which is surely somewhat different from the „Format” and „Low-level format” options that a HDD has for its data. Erasing data from a flash memory drive can be performed instantly, which justifies the „flash” term.
Although I said that the first (flash-based) solid stat drive was made in 1991, there had also existed some precursors of this technology – in the 1950s there were some magnetic core memories and card capacitor read-only stores (abbreviated as CCROs). Then, some decades later, there were some SSD implementations for the first IBM, Amdahl, and Cray supercomputers, but with limited usage; as for another example, in 1978 Texas Data Systems issued a 16-KB solid state RAM drive.
It was 1995 when flash memory SSDs started to become more acknowledged, starting step-by-step to replace HDDs in the military and aerospace domains. And of course, more recently in the 2000s and till today, solid state drives based on the Flash principle developed more and more, getting to replace HDDs to a certain extent in our personal computers too.
Now, putting historical aspects apart, how do SSDs work?
Their flash memory is able to provide higher performance. Flash memory in SSDs uses the NAND technology, for this reason SSDs can be thought of as large USB drives. The NAND technology is based on some electronical stuff called floating-gate transistors, alternatedly storing binary zeros and ones, and these transistor gates are organized like a grid structure, which also fits into a block. There is also a SSD controller that, you guessed, controls the way that data is working in these blocks, either to write to the drive or to read from it.
A particularity of SSD blocks is that one given block has a limited number of times it supports being rewritten. In order to prevent early drive wear, the internal system of the SSD drive intelligently distributes the data to be written across all sectors so that re-writings of the same sector occur more rarely. Otherwise, the life span of the SSD drive would surely be shortened, since we would no longer be able to work by writing data to it.
Another important thing to know is that there are actually two different flash technologies, whose names are related to logical terms:
- NOR technology, where the grid-contained transistors (also called cells*) are wired in parallel;
- NAND technology mentioned above, differing from NOR not only by name, but those transistors are wired in series, which allows transistors to be packed on a chip in a greater density, using a smaller number of wires – and this advantage of density explains why NAND is preferred when manufacturing SSDs.
*In a SSD structure, each two transistors make a cell, with one of the transistors working as a control gate and its “brother” as a floating gate. This is about how current circulates through transistors – the current arrives at the control-gate transistors and electrons flow further to the floating one, where this current flow is interrupted by a positive charge. Since the data we store on drives is digital (ones and zeros), the way that SSD transistors work must provide us with precise patterns of 1s and 0s for data accuracy. Remember that SSDs and HDDs are non-volatile memory for our data, they keep the data when the power is off.
NAND flashes are cheaper and much faster than NOR ones! Of course people would choose NAND over NOR, given these conditions.
But there is some further information on this NAND SSD technology! It is itself divided into several variants of implementation – like single-level cell and multi-level cell, where the SLC stores one bit per cell and the MLC stores two such bits. Although MLC wears out within a shorter time, it delivers higher capacity and is cheaper per gigabyte, which is why this is “the preferred of the preferred” implementation for SSDs. Single-level cell implementation is the fastest due to the single bit per cell, but its storage capacities are not very high.
More recently, there also emerged TLC (Triple-level cell tech, very common nowadays) and even quad-level cell, called QLC technology. While being slower due to storing 3 and respectively 4 bits per cell, they offer larger and larger, less expensive drives. The QLC tech, which is naturally the newest, is momentarily being used on just a few products like Intel NVME SSD 660p, Crucial P1 NVME SSD, Samsung 860 QVO SATA3 SSD.
Pros and cons in Solid state drive vs hard drive
The absence of moving and fragile parts like the mechanical head part of a HDD makes SSDs more durable. SSDs are made from semiconductor chips, whereas HDDs contain magnetic media. The fact of being made of chip components instead of mechanical moving parts is the reason behind the “solid state” name SSDs bear. Also, a SSD is supposed not to reach temperatures as high as a HDD can, while also requiring less energy (watts per hour).
Another benefit of the flash memory and the NAND technology that it relies on is that write and read speeds are higher – SATA III hard disks are able to make us enjoy with transfer speeds of 100-300 MB/s, but SATA SSDs may give out 600 MB/s speeds, that is clearly an improvement over hard disks. And SATA-III is NOT the only standard SSDs are sold nowadays, we will run into some other hot details later.
As for the failure rates, there is a parameter called mean time between failures or MTBF, which indicates the reliability of the HDD or SSD, and in the case of SSDs it may be up to five times higher, meaning SSDs are more durable, also because there are no moving parts like spinners and platters inside a solid state drive. We may see a HDD MTBF of 500000 hours, or a range between 300000 and 1200000 hours (however, let’s not believe that this means HDDs can run for more than a century, this is about extrapolated statistics data, and those of you who have statistical knowledge will surely understand what this means, and what methods are used in that activity field), but as for a suggestive comparison, in the SSD case we may see MTBF values like 1.5 to 2.5 million hours, which clearly indicate that SSDs would not fail as soon as a hard disk is “in danger” to do so! From this point of view, SSDs are faster & safer… also more long-living.
Back to read/write data speed, other possible rates that we may see would be 50-150 MB/s for a SATA hard disk and 510-550 MB/s for a SATA SSD, and this way SSDs score again higher than HDDs. Then let’s remember that a hard disk contains noisy mechanical parts, while a SSD is quieter because, for example, there are no 7200 rpm-rotating platters, nor read/write heads powered by a mechanical motor.
However, as there is no perfect thing in our lives, this technological progress of SSDs also comes with some downsides that – starting from some point on – makes HDDs more preferrable. First of all, although their tech evolvement during the last years led to price decreasing, SSDs still cost more dollars per gigabyte/terabyte when compared to HDDs, they are still pricey, so their speed advantage comes with additional costs. As of November 2019, if we look for a 1 TB SATA III SSD, prices can range between $110 and $170, whereas a 1 TB SATA III HDD can cost from $30 up to $60 depending on the model. I can also make some calendar-based comparison – back in 2015, a SATA SSD was about eight times more expensive than a HDD, and as of 2019 the rate decreased from 8x down to 5x or even smaller as seen above, but in most cases we are still facing prices more than 3x higher for SSDs.
Then I was saying something about how many times SSD blocks can be re-written. When a bit of information has to be written on a SSD, large blocks of data have to be erased and rewritten for that, which means that erasing cycles are triggered on the two-transistor cells lying in that area of the SSD where the involved data is. During the erase cycle, the floating-gate transistor is electrically charged and its resistance increases, which means that by time it becomes more difficult to flip this gate for doing writings and more current is needed – so that at some time in the future, the floating gate cannot move any longer and that SSD zone becomes read-only. So no more new data written. On the contrary, reading only needs checking the voltages of these cells, without modifying them, so we could continue to use that “writing-decayed” SSD part for reading data all right, but… of course our usage would be impaired by the read-only phenomenon.
Some specialists that noticed this decaying of the writing performance managed to find a temporary solution for delaying the decaying (rhyme intended) by distributing SSD writings over all the blocks. This technique is called wear-leveling. However, after a certain time of steady writing work, the wearing (decaying) of the SSDs will eventually occur. And here comes a quality difference between the single-level cell NAND technology and its “multi” counterpart: SLC typically offers us 50000 program/erase cycles, but the more available MLC decays after about only 5000 such cycles. This is why some people prefer to alternatedly use a SSD (for the operating system that loads faster, in a matter of seconds) and a HDD (or more, why not?) for the data they use frequently / on a daily basis, like photos, music, movies, content creation etc., making efforts to find the best balanced solution between a not-so-frequently-written-on SSD (faster) and a more-writable HDD (that is however slower).
In addition, not only SSDs keep on being several “x” times more expensive per gigabyte than HDDs, but their storage capacities are also numerically lower – earlier in the post I was talking about 16 TB HDDs and 4 TB SSDs. Well, as of November 2019, these are the upper storage limits for these drives. And 4 TB SSDs are pretty rare, with their prices being ranged somewhat between $500 and $1000, depending also on having discounts / being sponsored or not. 16TB HDDs are also new and still not largely available, but their prices are somewhat similar, like $500-600 and also depending on promotions, but they are four times larger! And for 2020 we should expect even 18 TB pieces. So, for large storage data centers, HDDs still have a future, and the time for SSDs has not come yet.
Or, if you need tens or even thousands of terabytes (maybe even more…) of storage for video surveillance, or very huge and intensive content creation purposes, then large hard disks are by far the best available solution for your business. Surely there are still years ahead until SSDs will come in 16 or more TB fashions, and with enough affordable prices for purposes of those kinds.
Solid state & hard disk drives are not only SATA!
Till now I focused the article on the widely known SATA III technology for HDDs and SSDs. But there are also other form factories such as Thunderbolt (version 3 nowadays), USB (current version: 3.1) and – this time only for SSDs, here comes a real piece of cake – the fastest PCI-E NVME SSDs! There are also SATA III compatible SSDs that have a special form factor, similar to NVMEs, and for both PCI-E NVME and SATA III drives mentioned here there’s a name: M.2 .
When it comes to Thunderbolt solid state drive vs hard disk drive, if we talk about write speed, the SSD can do it at over 400 MB/s, but the hard disk counterpart stays between definately lower limits like 60-110 MB/s, depending also on how much of the disk drive space is already used.
USB and Thunderbolt HDDs & SSDs can be used as external disk drives, where SSDs are more shock-resistant and less heavier than HDDs, their capacities range between 120 GB and 2 TB, and there are also some models (like the Samsung Portable SSD X5 2 TB) that while supporting Thunderbolt 3, they make use of the NVME power for read/write speeds of up to over 2 gigabytes per second (yes, that is it), which means that SSDs can be up to 25 times faster than HDDs when it comes to external drives. A bit later we will talk about NVME too. USB SSDs can reach and even get over the 500 MB/s data transfer threshold.
On the other hand, USB & Thunderbolt HDDs, that are used as external drives meaning they are portable, are able to provide us with insanely high storage space. There are some devices like the LaCiE 2big Dock Thunderbolt 3 suite, coming in 20 TB and 28 TB variants and being able to benefit from both Thunderbolt 3 and USB 3 ports. However, these devices are suitable for Mac computers that are relatively expensive, besides the disk drives being themselves expensive (like over $1500 for the 28 TB part).
USB hard disk drives can have data speeds between 100 and 200 MB per second.
And now let’s focus a bit on the smallest and (for SSDs) the fastest disk drive factor form of these days: the M.2.
M.2 devices can have either a SATA 3 or a PCI-E connection, but each of them can be plugged into a small port from the motherboard (called exactly M.2 port), modern desktop motherboards feature up to four such M.2 ports, not to mention PCI-E expansion cards that can be connected to the PCI-E slots and provide extra M.2 ports for storage. The M.2 drives are physically small, and they regularly require a special heatsink above them in order to dissipate the heat (because they get warm, they are a special species of SSDs). There are also people like ModMyMods who sell waterblocks for cooling M.2 devices, so we have to admit that today’s technology is just amazing!
And let’s be clear: there are no M.2 hard disk drives, here we are talking only about SSDs and we will see which of them are the strongest – hereby I mean the speediest M.2 SSDs. SATA M.2 SSDs can reach 600 MB per second, which makes them similar to other SATA 3 drives in matter of transfer performances, but of course they are expensive. Most M.2 drives are 80 millimeters long and 22 mm wide, hence the 2280 standard.
Now, like saving the best for last, here comes the most exciting part: the NVME PCI-E SSDs. These slim storage drives, that are also among the most expensive SSDs in the world, deliver much higher read/write speeds when compared to SATA and USB, because they directly benefit from the power of the PCI-E lanes from the motherboard… just like the cards we plug into the regular PCI-E slots (video, network, expansion…). They are so small (like the 22*80 mm form factor) and so fast, with the best PCI-E 3.0 parts being able to perform 3500 MB/s when reading and kind of 3200 MB/s while writing – yes, writing is more difficult, remember that it also requires modifications on the cells’ voltages – and we are not finished with these speeds: the PCI-E 4.0 standard has recently come out along the AMD X570 chipset that some new AM4-socketed motherboards belong to (and there are other PCI-E 4.0 AMD motherboards to come…), and now getting back to NVMEs, there are pieces of cake like the new Sabrent 1TB Rocket NVME 4.0, which with the power of the new PCI-E 4.0 standard is able to give out read speeds of 5000 MB/s and write speed rates of 4400 MB/s. Isn’t that one very fast!?
That one can be looked for and subsequently found out on the Internet having a price about five times higher than an average 1TB SATA-III HDD. There are also higher-capacity, extremely speedy NVME SSDs like the Corsair Force Series MP600 2TB Gen4 PCIe X4 NVMe M.2 SSD, featuring 4950 MB/s for read and 4250 for write, and its price is well above $400 US, however still affordable for a PC enthusiast that has generous monthly earnings and is keen on using strong hardware pieces.
Maybe I am wrong, but it looks like these newer NVMEs are somewhat cheaper than their PCI-E 3.0 counterparts… Technology makes progress!
There are also other SSD-related solutions like the Add-in card (it plugs into a PCI-E x4 or x16 slot, like a video card, and benefits directly from the PCI-E powers, and Intel Optane memories are such examples) and the U.2 2.5-inch drives, that use a specialized U.2 connector on the motherboard (assuming that the motherboard has at least a U.2 port), benefitting also from the PCI-E interface and tending to be more expensive & larger in capacity than average M.2 drives. U.2s are suitable for servers.
A bit on hybrid drives
There are also a particular kind of storage devices that was implemented in order to benefit from both kinds of storage drives while using a single piece. They have rotating HDD platters and also some area of NAND flash memory placed into the same object, that is called a hybrid drive. The data is read from the “hard” part, then the most used bits are stored into the “memory” area, basing on the Cache memory principle. Of course the data from the NAND is subjected to change during the time, but the fact of storing the most frequently (“favorite”) bits into NAND will provide that disk drive with SSD speeds. It is stated that hybrid drives are (slightly) more expensive than regular HDDs, but substantially cheaper than capacity-equivalent SSDs, since the amount of the “solid state content” inside a hybrid is pretty small, kind of 8 GB of flash memory along with 2 TB of platter (HDD) storage space. Yet we are talking about a single disk drive!
When starting activity on such a drive, or while writing new data, it will however perform as slow as a regular hard disk drive, until the caching is done. Then, by the time, the speed of use improves as much as caching gets done. However, given the fact that pure SSDs decrease in price, hybrids are supposed to lose their reason of existence. Still, it is at least interesting to learn about them.
As we can see by patiently reading through the post, solid state drives and hard disk drives come with specific features, pros and cons for each side, and especially SSDs have already an entire variety of showing up on the market (2.5-inch, 22*80 mm, cards / SATA, PCI-E, USB, Thunderbolt ports). Speaking in terms of price-per-storage-unit, hard disks still dominate the world, but if we want speed supremacy, this one comes to SSDs, and even to small SSDs. The ongoing history of storage drives meets also the intermediary category of Hybrid drives, which have their usefulness as well. When choosing the storage space for our computers, we have to make it clear about what our purposes of use are and about our available budget too, while also looking for the components that would best fit our needs, and knowing what sort of performance they are able to provide.