SSD Common Terms and Why Enterprise-Level SSDs Cost More
SSDs, also known as Solid-State Drives, are a type of persistent storage device based on programming information to flash chips within larger cells. Currently, the most common form of storage is HDD, the traditional Hard Disk Drive, which relies on physically programming information onto the machine.
Why are Solid-State Storage Devices So Popular?
SanDisk introduced the first SSD for commercial purchase starting in 1991 and immediately made a huge impact on tech corporations looking for more reliable, faster, and more compact ways to store large amounts of data.
In order to understand why solids state drives are so popular, it's vital to first explore the alternative option of storing information persistently: hard disk drives. These drives are composed of several pieces: the platter, the track, and a disk head. A platter is the circular, hard surface where data is stored by inducing magnetic changes to its surface. The disk might be composed of two or more platters, and each side can be programmed for reads and writes to memory. Even when the computer is powered off, the device retains its magnetic charge.
The platters are bound together around a spindle which rotates them several thousand times every minute. To get information from the disk surface, a disk arm - attached to a sensitive disk head - positions itself over the platter to read or write information. Information is stored on addressed disk tracks so that the disk head knows where to land when the client requests specific data.
Due to their design, HDDs are slower to make changes and process information. As our hardware gets increasingly small, it's harder for the disk head to accurately read and store information persistently. Our technology has reached the near limit of how fast the disk head can traverse over the surface of the platter to read information in.
By contrast, solid-state drives have no mechanical or moving parts. Instead, they are constructed out of transistors (a unit of semi-conductive material) organized into bigger blocks dubbed flash chips. Information is stored on these transistors by first erasing the pre-existing information on the chip, storing it on a dedicated scratch-work block, and then almost instantaneously re-programming it on the chip. SSDs are much more similar to RAM (Random Access Memory) in that they are "random access" devices, since the SSD can program information equally as fast to any component of the drive.
In terms of performance, SSDs are magnitudes faster than HDDs. While the average HDD can store one random request at the rate of 2 megabytes per second, a mainstream solid-state drive could fulfill the same request at 287 megabytes per second. They also perform better on metrics of reliability: a flash chip can be programmed over 100,000 times without losing information, whereas a hard disk drive block can be programmed accurately only 10,000 times.
However, solid-state drives also cost nearly ten times more than the traditional HDD. Despite this expense, most businesses agree that solid-state drives are the way of the future and more than worth the cost.
Enterprise-Level and Client Drives
One of the most important distinctions between drives is the difference between enterprise versus client SSD. These distinct types of drives have more than a magnitude of difference between their relative measurements on reliability, speed, and performance.
Client drives - also known as mainstream drives - are often smaller, single-cell, and less reliable. They are geared to work well with read-intensive workloads; hence, the number of write operations that can be completed per day is limited. Since most applications don't require a high volume workload, many businesses have adopted the use of these client drives. However, it's vital that companies that project a high throughput, or require large and frequent changes to persistent memory, invest in the best enterprise storage device with the highest reliability, speed, and endurance.
Additionally, client drives have less over-provisioning than enterprise-level drives. By the nature of how solid-state drives are written to, a large chunk of information needs to first be erased and then reprogrammed whenever old data is changed. Over time, this reprogramming process can strip the unit of the ability to store new information. Enterprise drives have a larger degree of over-provisioning which decreases the space and temporal cost of changing information in the drive (known as lowering write amplification) in addition to improving the reliability of the whole drive.
Enterprise drives reserve some space on the flash chips to copy old data, thus increasing reliability, whereas client drives use virtually all chips for storage. Therefore, enterprise-level drives have greater endurance and consume relatively less power per information transaction which is critical for high data-throughput workloads.
Common Solid-State Drive Terms
In order to make the best purchasing decision, it's important to understand the different acronyms surrounding solid-state drives. Knowing these key terms will help every entrepreneur make the most informed decision on the type of persistent storage device they should invest in.
Different Cell Levels:
SC - Single Level Cell, MLC - Multi level Cell, TLC - Triple Level Cell
Flash chips are meant to store one or more units of information on a single transistor, where the degree of magnetic charge determines the binary value of the transistor at that location.
A single level cell (SLC) means that only one bit - that is, either a "0" or a "1" - is stored inside a transistor. Due to their simplicity, SLC flash memory is the fastest and most robust memory available on the market. The firmware doesn't need to perform complex calculations to find the data stored in the flash chips, so there's very little overhead cost for these chips.
Multi-level cells, by contrast, store two bits of different levels of charge in one transistor. As you can imagine, a triple-level cell stores three bits of information. Triple-level cells are found in most personal drives as they are the least expensive yet are also the lowest performance, suitable for light client workloads or to be mostly read from rather than written.
RI - Read Intensive, WI - Write Intensive
A read-intensive drive focuses on increasing the performance of reads. When a client opens a file yet makes no changes, the drive is optimized to quickly retrieve that information for the user. By comparison, write-intensive drives are better for enterprises that focus on changing high volumes of data regularly, such as keeping track of online transactions, computing, and data warehousing.
Mixed-use drives perform relatively well on both reading and writing data yet aren't quite as well suited for either task. It's important to keep these concepts in mind because a business' storage devices should be aligned with the type of workload they expect.
PLP - Power Loss Protection
Power Loss Protection is a technique used to ensure the safety of the drive during and after power loss events. This security technique makes sure to flush out data that is traversing from RAM, such as data that has not been entirely transferred to disk yet and maintains the integrity of the drive's address table so that the location of information is always known. Often, large banks storing sensitive client data rely on PLP to make sure no information goes missing in an unexpected event.
DWPD - Drive Writes Per Day
VE - Value Endurance, HE - High Endurance, ME - Mixed or Mainstream Endurance
Endurance is a concept closely linked to DWPD which is measured in terabytes of transferred information per day. Value endurance (VE) is the cheapest option which indicates that a drive has one unit of DWPD, and generally means that these drives perform well on reading but poorly on writes. Mixed or mainstream endurance (ME) is sufficient for businesses that store small amounts of information and performs on a factor of 3 DWPD. By contrast, high endurance (HE) almost guarantees a performance of 10 to 25 DWPD and maximizes read and write performance.
EB - Enterprise Boot
Enterprise boot drives come in capacities of 60 to 120 gigabytes and include an Endurance Manager which helps to maintain the endurance for boot or swap applications. These drives perform well on low workloads for both reading and writing information and promise strong reliability.
Digitally Signed Firmware: DS
Digitally Signed firmware is a vital security measure built into drives to make sure that data theft, destruction or manipulation of data, and counterfeit device interactions do not occur. This firmware (FW) performs validity checks to authenticate that the drive and its input/output devices haven't been altered by third parties.
SC - Smart Drive Carrier
SC Converter – Smart Drive Carrier with Converter
Smart drive carriers are compatible, external devices meant to proactively warn against data loss when removing parts of the drive. The converter facilitates the connection of external devices to the solid-state drive, such as SATA, mSATA, or other peripheral devices. The converter also has a separate firmware controller that makes sure that slower devices don't become a performance bottleneck for the whole system.
IOPS: Input/Output Operations Per Second
Input/output operations per second (IOPS) is a performance measurement for storage devices. It refers to the maximum number of reads and writes to non-sequential storage locations on the drive. For HDDs, this measure is comparatively much lower than solid state drives, since solid state can program any chip equally as fast regardless of its location. Solid state drives maximize IOPS measurements.
While enterprise SSDs are more expensive than historic hard drives, most businesses agree that they are worth the cost in terms of their increased reliability, speed, and data persistence guarantees.