HDD, SSD, NVMe - what's the difference and which one to choose?
Basic disk information
What is a disk drive and what is it used for?
The disk drive is one of the basic components of a computer and one of the two responsible for storing data. The disk stores permanent information, such as software (including the operating system data) or user's documents. The second component responsible for data storage is the RAM, and unlike the disk, it holds temporary data, such as an edited, unsaved project or system processes. However, while data stored in RAM is lost when the power is turned off, data stored in the disk remains available even when the computer is powered down.
The disk may seem essential for your computer - but actually, its presence is not necessary for the machine to work. There are operating systems that can be run from removable media such as discs or flash drives. For the simplest ones, the now archaic floppy disks will suffice. Nevertheless, most computer users will use standard systems installed on disks.
Back when hard disks were not available for small computers or were too expensive for most users, floppy disks were the main storage medium. This is why, to this day, Windows drives names start with C: instead of A: - the first two letters were reserved for the floppy drives, as they were introduced earlier. Although they are no longer used today, this standard is still retained to maintain compatibility with older software
Disk drives types
There are two types of disk drives used in modern computers: HDDs and SSDs described below. Each of them has its pros, cons, and applications. In the case of the latter, we should also distinguish NVMe drives - plugged directly into the motherboard and several times faster than the others.
HDD
HDD - which stands for hard disk drive - is, along with the SSD, one of the two types of drives found in modern computers. For a long time, they were the main way for most users to store data.
A platter with hollow tracks holding binary data - zeros and ones is a HDD's main component. Special heads are used for writing and reading operations, they either read magnetic recording on the disk or overwrite it. As the disk operates, they move up and down while the platter spins at a speed of thousands of revolutions per minute - the more revolutions per minute, the faster the hard drive. Modern drives spin at a speed of 7,200 revolutions per minute, slower ones - 5,400. "Record-holders", described in a later section, reach about 15 thousand revolutions.
Unfortunately, this presence of moving parts causes the biggest drawbacks of HDDs: their data access times are much longer than SSDs'. It also makes hard disks highly vulnerable to psychical damage, like shock or vibration. When a working disk is strongly shaken, the head can hit the platter rotating, and while it may not cause much damage, it can also destroy the entire drive. On the other hand, if the HDD has been shut down properly, the head will not be above the platter, but next to it, so the chances of damage are much lower.
Early HDDs were noisy, making distinctive noises, most noticeably during intensive operations such as writing large amounts of data. Today's HDDs are virtually inaudible.
Hard disk drives used to require a lot of cabling and an external controller. Controllers integrated into the drives came along with the debut of the ATA interface. Over time, the ATA interface has evolved and later was superseded by the more efficient and space-saving SATA - Serial ATA, which is still the interface used by HDDs and some SSDs today.
Early data storage and HDD history
Magnetic forms of data storage were in use even before hard drives were invented. For example, the UNIVAC 1 computer that was developed in the early 1950s stored data on magnetic tape. The first HDDs appeared a few years later, in 1956 when IBM's 305 RAMAC was introduced. The hard disk of this machine consisted of 50 platters, each 61 centimeters in diameter, and was only able to store about 5 megabytes of data.
In addition to platters, those disks also had several mechanisms and heads. They took up a lot of space - they needed more space than an entire modern computer, as they were kept in "cabinets".
The first "compact" disk drives, with a diameter of 5.25 inches, appeared in the '80s. Along with the smaller size also came smaller capacity - they could barely store as much data as the REMAC's drive did, even though it was two decades older.
However, compact hard disks quickly "caught up" and then "outgrew" their larger predecessors - just a decade later, the largest drives could hold around a gigabyte of data. Today, HDDs usually have between 1 and 6 TB of capacity, and at this moment the "record holders" can store almost 20 terabytes of data. It is estimated that by 2025 they will reach the milestone of 100 TB disk space.
Should you use a HDD?
As the technology evolved and changed, the role of the "main" drive - the one containing the operating system and most important programs - has been taken over by much faster SSDs, described below. Despite that, the HDDs have an advantage in another aspect, the price-performance ratio: for less money, you can buy a drive with much larger disk space. For this reason, HDDs are most often used to store and archive huge amounts of data, such as documents, movies, or projects. Their other advantage is their longevity: these drives often exceed the manufacturer's stated lifespan. There are retro enthusiasts are owning decades-old computers with just as old HDDs that still work, although their low capacity and slow read/write speed render them useless by modern standards.
At the same time, it should be remembered that these drives are extremely sensitive to any physical damage, especially during data writing and reading when the head is close to the platter. For this reason, HDDs should not be used in laptops, which are particularly vulnerable to shocks and falls. While in newer laptops SSD drives are installed (with NVMes being installed in the more expensive ones), many older ones are still equipped with a HDD. In this case, the best solution is to replace the old drive with a resistant SSD, which is not only a safer option but also makes the computer significantly faster.
SSD
SSD stands for a solid-state drive. This name refers to the lack of moving parts, which distinguishes this type of drive from HDDs, and to the usage of transistors instead of electron tubes. Previously, SSDs used the same interface as HDDs - SATA. Today, the faster ones use the M.2 slot, and the fastest ones communicate using the NVMe protocol.
SSDs appeared on the consumer market relatively recently - they reached the mass market in the last decade. At first, they were perceived rather as a technological curiosity: despite their many advantages, they were much more expensive while less capacious than HDDs. Within a few years, however, due to technological advancement, they gained more space and their cost decreased significantly, making them affordable for home users. As mentioned before, unlike HDDs, SSDs have no moving parts, which makes them quite resistant to physical damage, as well as completely silent. They are based on cooperating semiconductor memory chips, which, in turn, can be divided into binary data storage cells. The drive has two main sets of memory chips that work together, seemingly to a RAID 0.
You can read more about RAID here.
Depending on how many data SSD's single memory cell can store, it will be categorized as one of those types:
SLC - single-level cell - single-level cell - can store only one bit, a value of 0 or 1;
MLC - multi-level cell, multi reflecting that it can store more than one bit - it can store two bits, values between 0 and 3;
TLC - triple-level cell - a single cell can store three bits or values between 0 and 7;
QLC - quad-level cell - it can store four bits, with possible values being between 0 and 16;
PLC - Penta-level cell - up to five bits can be stored, meaning values range between 0 and 31. PLC-type SSDs are not yet available on the mass market.
A cell's lifespan is inversely proportional to its capacity: cells that hold fewer bits last longer than their capacious cousins, and also are faster than them. At the same time, SLC and MLC SSDs are significantly more expensive than TLC ones.
Most users choose between QLC, TLC, and MLC drive. The first two have a lower price and larger storage space, while MLC has better performance and longevity. SLC drives are the most expensive, but also the most efficient alternative. They are mostly used in industry, due to their temperature resistance - they remain functional within the range of -40 to +85 degrees Celsius, much more than HDDs. Due to the technological barrier, however, this does not apply to MLC, TLC, and other multi-level cell drives.
It's also worth knowing that SSDs' controllers contain algorithms that distribute files written on the disk in a way all cells work evenly - in other words, they ensure that all cells are in a similar state. This optimizes and extends the drive's lifespan.
On the other hand, SSDs often prove in tests to have a longer lifespan than the manufacturer declared. While it is a fact that they will stop functioning after a certain amount of data has been written, high-quality drives can even last for a dozen years or even more, given they are not extensively used. In theory, this would make it possible for them to outlive most of the other parts of the computer.
An important event in the history of drives of this type is a test conducted by Techreport.com between 2013 and 2015, which involved continuous data writing to disks to determine how much could be written between they would cease functioning. The first drive to die was a TLC one, but even before it broke down, it wrote about 800TB of data - far more than the declared. The three most durable wrote 2.5 PB (or 2 500 TB) of data each. With such endurance, under normal use, a disk can work for as long as 2-3 decades.
SSDs also have other advantages, not directly related to their performance: they work in lower temperatures than HDDs and also have lower power consumption than them - in other words, they are more ecological.
Recovering data from a damaged SSD is much more difficult than from an HDD, so it's not recommended to keep important data on them, at least not without a backup.
SSD history
Before SSDs in their current form were introduced, drives based on DRAM were sold. These were offered as replacements for hard drives, but they were not actual SSDs and communicated through memory interfaces.
One of the first - if not the first - drives based on solid-state technology was the StorageTek STC 4305, released in 1978, offered as a replacement for one of the hard drives offered by IBM. It was praised for greater speed than its competitor while costing half the price.
In 1980, Toshiba employee Fujio Masuoka developed the flash memory (which is the base of the SSD technology), which the company commercialized in 1987. The potential of this memory was also recognized by SanDisk, which released the first drive of this kind in 1991.
Another important product in the history of solid-state drives is the FFD - Fast Flash Disk - released by M-System in 1995. Like modern SSDs, it could handle extreme temperatures much better than platter drives and was resistant to shocks, bumps, and falls. At the same time, it was less capacious and much more expensive than HDD drives, nevertheless, it became used by the army, flight, and space industry. They are used, among others, in flight recorders (such as cockpit voice recorders, or CVRs), commonly known as black boxes.
Over the next decade, this technology developed in parallel with HDD technology, which in turn became faster and faster with the introduction of the SATA interface and its subsequent versions. At the same time, the NAND memory became both cheaper and more capacious. The first consumer SSDs appeared in 2010/2011, but at first, they were too expensive for the average user. However, as time passed, the prices decreased and the drives gained better and better parameters until they finally became a basic component of a modern computer.
Today, SSD pioneers are racing in achieving faster and faster writing/reading speed and bigger capacity. As 3D NAND evolves, these drives will decrease in size and increase in capacity.
The largest SSD, described below Nimbus Data's ExaDrive, managed to surpass the largest HDDs incapacity. In 2019, Gigabyte presented a drive that reached around 15 Gb of reading and writes per second. Samsung, on the other hand, presented a drive that automatically transfers data from bad sectors, making it able to work just fine despite the damage, albeit with a smaller capacity.
Should you use SSD?
For most users, the SSD will be the main (and often the only one) disk drive of the computer - thanks to their speed, they leave the HDD far behind. If the drive is not meant for storing important data - then SSD will be the right choice for you.
When using an SSD, keep in mind that it is more difficult to recover data after a failure than in the case of HDDs, so make backup copies of the important data you keep on it. You can store the backup on external drives or in a cloud.
NVMe disks
NVMes are a part of the SSD family. They are different by the way they communicate with the computer - they do not use the SATA connector, which with the increasing speed of SSDs has become too slow for them, but instead are directly plugged into the motherboard. This allows them to achieve much faster read and write speeds than drives using the SATA interface.
Which disk should you use on a laptop or a PC?
The differences and uses of HDDs and SSDs are described above. In this section, we will answer the question of which drive is best for you.
One of the main issues to consider is the needed storage space. Today, most home users need to store a few hundred gigabytes of data - the operating system, a few programs, and personal documents. Gamers need more - modern games can take tens or even hundreds of gigabytes. Additionally, since these games need fast access to data, gamers usually opt for faster SSDs. The situation is similar for those users who create graphics, models, or architectural designs on the computer - advanced programs, such as those for graphics editings like 3ds Max or Sony Vegas need fast access and a large disk.
Which disk should you use in a server?
A server with a simple website does not need much disk space and a few gigabytes are enough. More advanced servers that, for example, host cloud services or multiplayer games such as Minecraft or CS: GO have much higher needs, needing terabytes of disk space and more. Some servers have a total of tens of terabytes of disk space (they usually use HDDs), which are most often used as network drives.
For cryptocurrency mining, disk space doesn't matter as much as a computer's pure processing power. In the beginning, during the synchronization phase, data can be kept on an SSD for faster access times and then moved to an HDD; the exception is Chia, a space-based currency that needs terabytes of space. For this currency, HDDs are strongly recommended as they have more capacity for a lower price.
How to care for your drive?
To ensure a disk's long lifespan, you need to take proper care of it. It should be noted that maintaining an HDD is completely different than maintaining an SSD.
HDDs should be kept in a stable environment so that they are not subjected to any shock that might cause physical damage. This type of drive should also be regularly cleaned of unnecessary data and defragmented, i.e. the files scattered throughout the drive should be organized. Hard disk drives can suffer mechanical failures, such as head failure. Repair of such a disk can be carried out by a specialist under proper conditions (home repair methods will most likely destroy the disk).
This is not the case with SSDs: defragmentation is harmful to SSDs and should not be performed. Because of their design, writing data to them wears them out, so organizing them in this way only shortens their lifespan. Instead, its lifespan can be extended by limiting the amount of data written to it - larger files that do not need fast access times (for example, media downloaded from the web) can be placed on the HDD. In Windows, defragmentation is often enabled automatically, even on SSDs. Therefore, we recommend that you check this setting and, if enabled, disable it.
Why is my disk less capacious than manufacturer declared?
This is caused by the fact that humans use the decimal system, while computers operate on binary systems. Mega-, giga-, terabytes, etc. are bytes multiplied by thousands, millions, and billions, i.e., multiply of 10 that are easy to use for a human. Computers, on the other hand, use the number 2 and its power; units based on these values are called mebi-, gebi- and tebibytes. It can be written with an example like this:
1 kibibyte = 1024 bytes,
1 kilobyte = 1000 bytes,
Successively multiplying these numbers yields the following values:
1 tebibyte = 1024 GiB = 1 099 511 627 776 bytes,
1 terabyte = 1000 GB = 1 000 000 000 000 bytes.
Manufacturers provide the value in giga- and terabytes; computers, on the other hand, calculate these values using the binary system.
How do these values translate into disk capacity? Because:
1 GiB = 1.074 GB
1 GB = 0.931 GiB
we can easily convert the specified size in GBs to GiBs. Therefore, for example, a drive is sold as 512 GB will be 476.84 GiB in size; a 1000 GB drive will fit 931 GiB, and so on.
RAID
While browsing information on disk drives or network service offerings, you've probably come across the term "RAID array." A RAID consists of two or more connected drives that can share tasks for better performance or copy contained data for security depending on the type of the array. You can learn more about RAID arrays here.
Disk file systems
Each disk stores files following a specific method called a file system. While the operating system can choose the default file system itself, users can usually select one manually.
Windows mainly use the NTFS file system; occasionally smaller external drives use its predecessor, FAT32, which was developed for older systems. Linux systems use many filesystems, for example, XFS, BTRFS, and the ext* family.
It's worth mentioning that most Linux distributions have built-in support for drives using NTFS, while Windows requires additional programs to read Linux partitions.
Comparisons
Fastest HDDs
In the case of HDD, speed can be understood in two ways: as the file write/read speed and as the number of revolutions per minute. While the most common are disks with platters spinning at the speed of a few thousand revolutions per minute, the record holders in this category spin even time 15,000 per minute. Such drives are produced by Seagate (called Savvio 15K), Dell, or IBM.
The fastest drive in terms of reading and writing data is Mach.2 from Seagate. According to its manufacturer, it reaches a transfer rate of 524 Mb per second.
Fastest SSDs
The fastest of the SSD family are certainly the NVMe. 2021's record holders are drives like the Samsung 980 Pro, the WD Black SN850, or the Sabrent Rocket 4 Plus, which read data with a speed of about 7000 Mb per second and write 5000 Mb per second.
Largest HDDs
Currently, the most capacious HDDs on the market can store about 18 TB of data - such models are for example offered by Western Digital and Seagate. The latter, however, has announced the release of a 20 TB drive. It is worth remembering that such large drives are not the fastest and instead are used for archiving - ordinary users will be satisfied with smaller ones.
Largest SSDs
While it may be surprising, there are SSDs capable of storing tens of terabytes - the record holder has a disks space of 100TB! These are ExaDrive drives from NimbusData. For most users, however, it will remain but a technical curiosity, as its price is 40 thousand dollars.
The largest consumer SSDs have about 8 TB of disk space.
Fun fact: longest operating drives
Due to their relatively short presence in the market, there is no long-lasting SSDs yet. However, in the case of HDDs, some disks still work 20-30 years after the first startup (of course, their efficiency is incomparably worse than their modern counterparts).
An interesting case is the Netware server described by the user Axatax on the arstechnica.com forum: started up in 1996, it worked without a break for 16 and a half years, or 6030 days, until it was finally shut down in 2013. It had two hard drives of 800 MB capacity each - together they worked for over 144 thousand hours!
Disks in Mevspace's (formerly Skynode) servers
The servers offered by Mevspace have both SSD and HDD drives. The fastest of them, the NVMes, offer capacities starting from several hundred gigabytes. On the other hand, the most capacious configurations have up to a dozen multi-terabyte HDDs (for example, 12 x 6 TB), which makes them ideal for backup or storage servers.
Summary
Both types of drives have their pros, cons and uses. SSDs are incomparably more efficient than disk drives, while the latter are less expensive while offering much more storage space. Thus, you should pick the drive after considering your needs.
Most home users will be fine with a SSD; the exception being those who, for example, store data like family photos or videos - they will need the large storage space of a HDD. On the contrary, gamers and those needing quick response times will appreciate the speed of SSDs. The most expensive option combining the use of SSD - as a system drive - and HDD - as storage - will certainly meet the needs of both basic and more advanced users.