Solid state drive (SSD) has been around for a while and its popularity is dramatically increasing among all classes of users. Most of these users only recognize solid state drive as a viable alternative to the traditional hard disk drive (HDD) with a much higher speed, and that very basic knowledge justifies for them to pay the extra cost and obtain one in hope to extend the performance capability of their systems.
We have added previously a few educational articles on our website to help those who are new to the world of SSDs make the right decision of purchase by clarifying issues such as: The Benefits A Solid State Drive (SSD) Provides To Computers, “Is Solid State Drive (SSD) Worth It After All?“, “How To Choose The Right SSD For Your Computer” and “Is 128GB Solid State Drive (SSD) Enough For You?“. And this article will just add to this educational series and help you understand the most common and important terms used in the SSD industry to enlighten you more about the thing you are going to buy (i.e. An SSD).
If you’re shopping for a solid-state drive—whether as a new boot drive or as an access-speeding cache for an existing boot hard drive–you’re likely tech-savvy enough to dig into the innnards of your desktop or laptop. Even so, a swarm of ever-evolving jargon buzzes around SSDs, and some of it is bewildering even to serious PC enthusiasts. Not only that, but not every spec that SSD vendors cite is necessarily meaningful when you’re shopping.
It’s hard to buy a bad SSD these days for general use, but first-time upgraders will need a bit of background knowledge to keep from overspending. Let us be your guide: Here’s a 101-level primer to the language you need to speak SSD-savvy.
Firmware refers to the software “instruction set” stored in an SSD in non-volatile memory. In a nutshell, it governs the operation of the drive. Firmware in an SSD context is referred to by a version number, and is flash-upgradable, usually via a manufacturer utility. The firmware is typically tied to a specific make and model of controller, so updates to the firmware for a given SSD controller chip can often be implemented across multiple manufacturers’ drives, as soon as each manufacturer packages the firmware update for its drives. Firmware upgrades are typically distributed via the support section of an SSD manufacturer’s Web site.
A firmware update can address performance issues with a given drive. Also note that a drive that has been on the market for some time may have shipped with an earlier version of a given controller’s firmware early on, and a newer one later, meaning that performance or stability can vary depending on which particular sample you buy.
2) Smart Response Technology (SRT)
SRT is an Intel technology that lets you install a low-capacity solid-state drive as a high-speed cache for a standard platter hard drive. It debuted with Intel’s Z68 chipset, and to implement it, you’ll need a compatible Intel-based PC, along with any SSD and hard drive. With SRT active, the system gradually “learns” which files and system elements you use the most, caching those to the SSD for faster access. In that way, you can get the advantage of the inexpensive high capacity of a conventional hard drive along with some of the access speed of an SSD.
Implementing SRT makes sense if you already have a hard drive in place as a boot drive and don’t want to go to the trouble of making an SSD your boot drive. That said, SSDs at capacities of 128GB and 256GB have gotten so cheap nowadays that there’s less incentive to do SRT for cost reasons; those capacities are big enough as boot and program drives for most buyers. And depending on how your system is configured, you may need to reinstall Windows on your hard drive, in any case, to configure things properly for SRT.
4) Serial ATA
Serial ATA, often abbreviated to SATA, is the standard interface for drives inside consumer and business PCs. It’s employed by hard drives, SSDs, and optical drives alike. Drives with a SATA interface will have both a SATA data connector (which connects, in a desktop, to one of the SATA ports on the motherboard) and a wider, blade-like “SATA-style” power connector (which connects to a SATA power lead coming from the power supply). Inside a laptop, these connectors on the drive usually engage with a hardwired connection or a very short ribbon cable with both connectors on it.
While SSDs do come in other interfaces and designs, the SATA SSD in its 2.5-inch form factor is by far the most familiar to upgraders. The SATA interface itself has speed grades, and the ones you’ll see in any SSDs you’re considering are SATA 2 and SATA 3, variously called “SATA II”/“SATA 3Gbps” or “SATA III”/“SATA 6Gbps,” respectively. These indicate the maximum data transfer rate possible with the drive, assuming it’s installed in a PC with a SATA interface supporting the same standard.
In current drives, SATA III/SATA 6Gbps is the standard; we mention this in the event you’re shopping older, second-hand, or remaindered drives that might be 3Gbps only. To gain the maximum throughput benefit of SATA 6Gbps, a 6Gbps SSD must be connected to a 6Gbps-compatible SATA port. Connected to a SATA II port, it will work, but the maximum data transfer rate will be constrained to 3Gbps.
mSATA defines both a form factor and an interface for compact SSDs. An mSATA SSD might be used as a boot drive (in a very compact laptop or tablet) or as an “SSD cache” (defined above), speeding up the operation of a mechanical hard drive by dynamically hosting frequently-accessed files or system/program elements.
An mSATA SSD is a bare circuit board, as opposed to the enclosed design of a 2.5-inch SSD. (It resembles, and is sometimes mistaken for, a Mini-PCI card.) It will have a blade-style data and power connector that plugs into a single mSATA slot. A few desktop motherboards of recent years have featured mSATA sockets on them, to allow for the onboard installation of an mSATA SSD for caching. In most cases, however, an mSATA SSD upgrade is mostly of interest to users of thin laptops looking to upgrade the mSATA boot drive in their machines. If you are interested in this type of SSDs you’re invited to visit our best mSATA SSD list where you can pick the best brands available on the market of this micro technology.
6) M.2 SATA
Until 2014 known more widely as NGFF (for Next Generation Form Factor), the M.2 standard is the successor to mSATA. Like mSATA, M.2 defines both the physical connector on the drive and the form factor, but with much greater variation for incorporating into mobile devices. Unlike mSATA, which complied with the dimensions of a Mini-PCI card, M.2 SSDs can come in a wide variety of sizes, from 12mm to 30mm wide and 16mm to 110mm long, and in versions with NAND chips on one or both sides. Though of lesser interest to aftermarket upgraders, this size variety allows makers of slim devices much greater flexibility for implementation. You’ll also see M.2 SSDs that work over the SATA bus, and others that will make use of PCI Express lanes, the latter allowing for greater theoretical throughput.
Indeed, with M.2, SSDs move ever further away from the packaged 2.5-inch form factor of conventional hard drives to allow for installation in ultracompact computers. We’re also seeing M.2 SATA slots on some of the very latest high-end motherboards featuring the Intel Z97 chipset, such as the Asus Z97I-Plus and MSI Z97 MPower Max AC, where the drive is intended for use in a caching arrangement. Note that M.2 is, in essence, a subset of the emerging SATA Express standard (defined later on), but it’s mostly meant for mobile.
7) Write Cycles
A longevity measure for SSDs, this spec (also called “program-erase cycles”) is more useful as a comparative attribute than as an absolute. It refers to the number of times a given memory cell on an SSD is likely to endure being erased and rewritten. (Typically, when a cell wears out, the drive decommissions it and activates another cell, if available, that’s kept in reserve via “overprovisioning.”)
In practical fact, most SSDs have proven to become obsolete in terms of capacity faster than their write limits are likely to be reached. You’ll tend to see higher write-cycle specs, however, for premium SSDs and drives destined for use in server or data center environments. These tend to be based on SLC, as opposed to MLC or TLC memory. (More on those terms later.)
8) TRIM Support
The way an SSD works, before you write to the drive, the SSD needs to erase fully any memory cells that are full of data to be trashed before it can overwrite them with new data. This becomes an issue especially once a drive starts to fill, and already-used cells are the only ones available for writes. If you’re doing this “maintenance work” at the same time as you’re trying to perform a data write, it can slow down performance.
Supported in Windows 7 and later, the TRIM command takes care of this chore in advance, looking ahead and pre-wiping available cells containing data to be deleted so they’re ready for writing when the time comes. Your SSD’s software utilities, as well as freeware like Crystal DiskInfo, can tell you if TRIM is activated.
9) RAPID Mode
RAPID Mode is a proprietary Samsung name for its SSD RAM-drive technology. It’s been included, so far, with its 840 EVO line of SSDs out of the box, and implemented via free download for some older Samsung SSDs, such as the SSD 840 Series. It stands for “Realtime Accelerated Processing of I/O Data,” and it works under Windows 7 and 8.
In it, a portion of your main system memory, which allows for faster access than even the flash memory on your SSD, is managed via a special driver to speed up data transfers. It does this by caching frequently accessed user data and application files. It can make benchmark performance extra snappy, but know that there’s a potential downside to RAPID Mode: Any power loss that occurs means that any data in the volatile RAM cache will be lost. (Remember: System memory needs to remain powered to retain its contents; the NAND chips in an SSD do not.)
10) NAND Flash
NAND flash is the generic term for the silicon chips that comprise the actual storage on the SSD. (The “NAND” refers, at a technical level, to the type of logic gates used in the underlying memory structure.) In essence, an SSD of whatever stripe is a circuit board with NAND chips embedded, managed by a controller (defined later in this story). This kind of memory is non-volatile, meaning that it does not require constant power to maintain the data stored on it.
The maker of the NAND on an SSD may or may not correspond to the actual brand of SSD. (For example, Samsung SSDs predictably will contain Samsung NAND, since the company also manufactures memory.) For the most part, the specific maker of the NAND is not a factor in an SSD purchase, though the kind of NAND (SLC, MLC, or TLC, defined below) might be, depending on how you will use your SSD.
11) SLC, MLC, and TLC NAND
These three memory types are the primary kinds of NAND chips seen in current SSDs. The most common are MLC (multi-level cell) and SLC (single-level cell). MLC is what you’ll see in most consumer SSDs; it’s generally the cheaper of the two. The “multi-level” of MLC refers to the ability of each MLC memory cell, in most cases, to host four states and thus two bits per cell due to its architecture. (SLC memory cells can only exist in two states, 1 and 0, and thus store one bit per cell.)
SLC in general is stabler over longer periods but also more expensive. MLC’s higher densities (which is to say, more chips out of a given wafer) make it cheaper to manufacture, but error compensation in the firmware is necessary to keep it in check. MLC also tends to be rated for fewer read/write cycles than SLC. A variant of MLC, enterprise MLC (eMLC), uses technologies that forestall cell wear and thus data loss, and premium-price drives based on these “stabler” drives are marketed for business or high-access environments.
TLC, meanwhile, is an emerging memory type used first by Samsung in its 840 Series SSDs, with other NAND makers also jumping on board. Standing for “triple-level cell,” TLC can host eight states and three bits per cell. The even greater density pushes cost down, but TLC requires even more error-correcting overhead, and the increased complexity and varying voltages per cell mean likely faster wear per cell, all else being equal. TLC has been seen so far mainly in consumer SSDs that won’t be subjected to mission-critical, enterprise workloads.
The silicon chip that acts as “traffic cop” for the SSD, the controller is typically the biggest differentiator among SSDs. Some manufacturers of SSDs have acquired controller makers over the years and incorporate those technologies into homegrown controllers (example: OCZ and Indilinx), while others make use of widely used controllers from Marvell or SandForce (the latter recently acquired by Seagate from LSI). Drives with the same onboard controller and of the same capacity tend to perform similarly, though different firmware versions and other factors can introduce variation.
13) Drive Z-Height
With a typical 2.5-inch SSD, the “z-height” refers to the thickness of the drive. Current SSDs come in two common z-heights: 7mm and 9.5mm. This doesn’t matter much for drives being installed in a desktop PC, which can accommodate drives of either height with ease, but for a laptop install, the z-height can be crucial.
Many recent thin-design laptops and ultrabooks that use 2.5-inch drives inside require a 7mm z-height drive to fit. (That said, increasingly, makers of extremely thin laptops will opt for completely different, thinner SSD form factors altogether, such as mSATA and M.2.) Some SSD makers will include a “spacer” (usually, a frame of plastic) with their 7mm models to help them fit securely in a laptop drive bay meant for a 9.5mm drive.
14) Migration Software
As a category, this is software that may or may not come packaged with an SSD to assist in copying a source drive to an SSD. (The most likely scenario in which it will be used is if you intend to install the SSD as a boot drive.) It’s not possible to simply copy a bootable hard drive to an SSD, bit by bit, within Windows, and have the SSD be bootable. Because this operation needs to happen outside of Windows, special software is required.
That said, the lack of migration software does not have to be a deal-killer; freeware like EaseUS’s Disk Copy can take its place. Some SSDs will supplement the migration software with a SATA-to-USB cable (for transferring the contents of a laptop drive over USB); when that’s included, the SSD is often marketed as a “laptop upgrade kit.”
Because memory cells fail over time as they are written and erased over and over, an SSD’s effective capacity can drop gradually as memory cells fall out of the running. Some makers of SSDs, to forestall this, provide more memory than advertised, or “overprovision” the drive.
You won’t be able to see this extra memory in the advertised capacity of the drive, or in normal use; the drive firmware may invisibly bring some of these cells online as other die. But it’s a sign that the SSD maker is factoring in gradual data-cell death. A secondary consideration: Overprovisioning means that the SSD can write to a wider range of cells, which proportionally reduces wear across the whole array.
16) Sequential and 4K Reads & Writes
The most common SSD benchmarking software programs, including the AS-SSD and CrystalDiskMark utilities that we use in our tests, typically test two kinds of data transfers: sequential reads/writes, and random (“4K”) reads/writes. Sequential reads and writes involve large files; testing in this fashion gives an idea of speeds when transferring large amounts of data. The term is a vestige of such operations on conventional hard drives, in which large files would often have most of their parts in a row, in physical proximity, on the actual drive platter.
Random reads and writes, on the other hand, access small (usually 4K in size) blocks of data, simulating the device saving and reading much smaller bits of data scattered across the drive. All of these measures are reported in megabytes per second (MBps or MB/second), higher being better. Note that when SSD vendors report claimed read and write speeds, they’re usually sequential numbers, both because the majority of data accesses on a client PC tend to be sequential, and because these numbers look the biggest. Some software and SSD makers report this kind of data in IOPS (input/output operations per second).
For “mean time between failures,” this is another spec that, if it’s meaningful at all when shopping, is only useful for comparison among drives from the same maker. It’s a measure of the expected rate of failures in a population of drives, and not as the projected absolute lifetime of any given drive in hours. (MTBF is often cited as a measure for other kinds of computer hardware, too, such as platter disk drives, but it’s only useful as a measure within hardware of its own type.)
A JEDEC standard outlines the testing of SSDs for longevity under reads and writes, but it’s not always clear if a given SSD vendor is using the same metrics and workloads to test for longevity as another. As a result, MTBFs are really only relevant for buyers if you’re looking at drives within the same manufacturers’ families.
18) Wear Leveling
Wear leveling is an internal management technique used by solid state drives’ firmware, to maximize the viability of all memory on the drive. In it, write and erase operations are spread across the entire drive, instead of concentrated on the same block of cells over and over, even if the drive is not filled to capacity. Doing so “wears” the cells across the drive evenly.
19) PCIe SSD
A less common (and significantly pricier) form of SSD, in this kind of solid-state drive the NAND chips are embedded on a PCI Express (PCIe) expansion card. The idea here is that the PCIe SSD bypasses the SATA interface and makes use of the wider bandwidth of the PCI Express bus. These kinds of drives, such as OCZ’s RevoDrive line, may be made redundant by SATA Express in coming years. They’re used mostly in servers or data-center applications, though you can put one in a consumer PC.
In 2014, we’re also seeing a variant on this kind of SSD from Plextor, in the form of an M.2 SSD mounted on a PCIe card in a socket, for desktop systems that do not have M.2 sockets.
20) SATA Express
SATA Express is not yet a shopping consideration in drives as of mid-2014, but it’s poised to become one in coming years. The first SATA Express-capable motherboards began to appear for PC desktops with the May 2014 wave of boards based on the Intel Z97 and H97 chipsets.
SATA Express is implemented via a dedicated connector on the motherboard that resembles an internal SATA port, but keyed differently. In essence, it employs the same principle as a PCIe SSD, in that the SSD makes use of PCI Express lanes for greater bandwidth.