Internal Flash Storage and Solid State Technology










PDF version of this report
You must have Adobe Acrobat reader to view, save, or print PDF files. The reader
is available for free
download
.

Internal Flash
Storage
and Solid State Technology

by Michael Gariffo

Docid: 00021029

Publication Date: 2104

Report Type: TUTORIAL

Preview

Flash-based storage hardware has been around in
one form or another since the 1980s. It was not until more than a decade after
its original release to the public that the technology began to flourish and
show real promise as an alternative to spinning disk drives or the older tape
drives that may still have been in use at the time. Lower power requirements,
lack of moving parts, and generally smaller footprint made it particularly ideal
for the portable device boom of the late 1990s. In fact, this surge in portable
products is what gave most consumers their first introduction to flash storage
technology, whether in the form of an MP3 player, as the storage card for a
digital camera, or as the onboard memory in a cellphone. While most tech-savvy
people owned one or more devices with flash-based storage by the turn of the
millennium, it would take nearly another decade before it reached its full
potential as an alternative storage drive in consumer-level desktop and laptop
PCs. This report will focus the history of flash storage’s growth as a desktop
storage platform, its current applications, and its advantages and disadvantages
when compared to traditional disk drives.

Report Contents:

Description
and History

[return to top of this report]

The Road to Becoming a Desktop Storage Solution

Flash storage saw its first
applications in the consumer and business product market as a storage medium for
portable devices. It excels in this area due to the fact that it does not need
continual power to retain its data as a more volatile memory type such as RAM (Random
Access Memory) does, nor does it need power to drive a spinning platter like a
hard drive would. This significantly reduces the power draw for mobile devices, allowing for longer run
times without the need for bulky batteries. Flash-based storage excelled in this
arena for many years, powering numerous portable storage mediums including SD
cards and their variants, Compact Flash cards, jump drives, and more. All of
these technologies offered a stable, relatively inexpensive way of providing
storage to portable devices that could be moved to and from a desktop PC or
laptop, without resulting in any significant power drain.

While portable devices were enjoying this newfound versatility, desktop and
laptop PCs were still stuck using hard disks as their primary (and often only)
method of storage. Smaller flash-based external storage solutions were
available, but they could not serve as the primary drive in a system due to the
technological limitations at the time. This was much more of an issue for laptop
owners as their portable PC’s battery would have to power the spinning platter
of a disk drive, putting a drain on its reserves whenever data needed to
be read from or written to system storage. Likewise, read and write speeds on devices
using disk drives could not match those found in flash-based storage due to the
fact that the platter on which the required information was stored would need to be spun into place before the read or write operation
could even begin. This physical requirement results in both a time delay and
additional draw on the system power for each read/write operation. On the
opposite side of the coin, flash-based storage solutions have no moving parts
and are able to access any and all data stored within their memory instantly. It
is this dichotomy that led technology manufacturers to begin exploring the
possibility of creating a full, desktop-class storage drive using only flash
memory.

Figure 1: A
Range Samsung Solid State Drive Form Factors

Source: Samsung

The Birth of Solid State Drives

Solid state drives (SSDs) first appeared as a storage
option exclusively for military and highly specialized scientific equipment.1
These early units were extremely cost prohibitive and had limited storage
capacity. In truth, they were generally designed to serve a single purpose within
a particular machine or weapon, with no usability beyond their original
installation point.

This began to change in 2007 when solid
state drives first appeared in consumer-grade products, including
commercial offerings such as a Dell ultra-portable laptop as well as the One Laptop Per Child initiative’s XO
Laptop.2,3 The latter was particularly well
suited to using an SSD, as it was expected to be used by children in developing
nations where rough treatment and harsh conditions would be the
norm. In such a case, hard disk drives would likely fail in short order. If a child dropped the unit while a read or write operation was taking
place, it could irreparably damage one of the drive’s storage platters, rendering
the entire laptop unusable. Similarly, most disk drives require what is called a
"breather hole." This opening is designed to allow in the air that hard drives
actually require to operate correctly, but can also serve as an entry point for
foreign matter.4 While these holes are
generally equipped with some type of filter to keep debris out, a
constant assault from sand or salty air could very easily circumvent this precaution
and cause the hard drive to fail.5
For contrast, most SSDs, even at this time, were and are completely hermetically
sealed, preventing any outside matter from entering the storage device and
protecting the user’s data as well as the drive’s continued operation.

While the aforementioned offerings showed the
possibilities of SSDs in consumer-class devices, it was Apple that brought the technology to the forefront of consumers’ minds by
integrating the option into a completely new laptop line, the MacBook Air.6
This laptop was considered almost comically thin at the time, with jokes being
made about how it could be used to slice a loaf of bread. The
unit’s dimensions and its almost unheard of battery life were due in large part
to a decision by Apple to offer a version with a solid state drive as its primary storage unit.
While a less expensive version was offered with a spinning drive
at its core, the battery life on the SSD model was markedly better with less
heat production and run times that exceeded the hard disk version by the better
part of an hour.7 Following the launch of the
MacBook Air, the flood gates for SSD-equipped laptops opened wide, with units
quickly following from nearly every major PC maker.

 It was in 2008 that the enterprise
segment got on board with the trend of integrating SSDs into its offerings. The
first solid state drive in an enterprise-class storage appliance was the EMC Symmetrix DMX system,
which used a STEC Inc. Zeus-IOPS SSD.8 Another important first in the
evolution of enterprise-grade SSDs came later that year when Sun released its
Sun Storage 7000 Unified Storage Systems line. This was the first storage
appliance to use a hybrid system that utilized both SSDs for faster
read and writes speeds when working on actively used data with hard disk drives
for storing larger amounts of data for the long term.9 This was a milestone for SSDs in several ways. First, it highlighted the strengths
and weaknesses of solid state storage, showing that, while it was much quicker
and more responsive than hard disks, it was also too cost-prohibitive to use
exclusively for
storing large amounts of data. Second, it showed that it was
possible to create a hybrid solution that leveraged the strengths of both
storage methods, something that eventually resulted in the creation of single
drives utilizing both technologies.10 Finally,
it truly introduced enterprise customers to the speed and efficiency benefits
made possible by SSD technology. Although the price per GB of storage in the SSD
market has come down from the lofty heights of 2008 for both consumer and
enterprise-grade drives, it remains very much in excess of what hard disk
storage of a similar volume can be had for. Because of this, the dual-technology
approach to storage remains just as important as it was a little over a decade ago.

Figure 2: A Seagate Momentous XT Hybrid SSD/HD with Its Internal Components Exposed

Figure 2: A Seagate Momentous XT Hybrid SSD/HD with Its Internal Components Exposed

Source: Seagate

Applications and Features

[return to top of this report]

Having tracked the path of flash storage and solid state
drive technology essentially to where it stands now, the next step is to take
a more in depth look at the SSD options, technologies, and varieties available
today. This section will focus on what SSDs excel at, where they fall short,
what connector technologies they currently utilize, and when they should be used
in place of the more established hard disk drive technology.

What Solid State Drives Excel At

This section focuses on the highlights of solid state drive
technology, describing areas where it beats out the capabilities of hard disk
drives.

  • Read/Write Speeds – Solid state drives
    are far and away superior to hard disk drives in this area. The fastest
    drives currently available feature sequential read speeds in excess of 7,000
    Megabytes per second (MBps), with write speeds of 5,000 MBps.11
    Although this is still somewhat faster than most consumer-grade models can
    expect to achieve, and highly dependent on the type of connector being used,
    even run-of-the-mill consumer-level drives typically offer read/write speeds
    in the 2,500 MBps range, with enterprise SSDs running in near lockstep.12
    For comparison, some of the fastest hard drives on the market today top out at
    less than 200MBps with even slower write speeds.13 It should be
    noted that this figure was obtained by a 15,000 RPM hard drive, which spins
    at more than twice the speed of an average, consumer-grade 7,200 RPM hard
    drive. It is the reliance of hard disk technology on spinning platters (and
    moving parts in general) that provides the storage format’s upper limit in
    read/write speeds. Hard disk drives can only spin so fast before they
    literally self-destruct. On the other hand, SSDs do not require moving
    parts, meaning the upper limit of their read/write speeds is defined solely
    by the electronic data transfer rate of the transmission technology being used.
  • Access Times – This aspect of storage
    drive operation is closely related to the one above and also differs
    greatly between the two technologies because of the need for moving parts in
    a hard drive. While a SSD typically needs less than 1 millisecond (ms)
    to access any information stored anywhere on its hardware, a hard drive can
    take up to 12 times that long to find information.
    14,15 This is a result of the mechanical components of the hard
    drive needing to move the platters and magnetic head to the correct
    location. While a difference of 11ms may seem trivial, this is the
    time it takes for the drive to locate and begin reading or writing each
    fragment of data within a given file. Since files can have segments spread
    across thousands of locations on a given drive, particularly if that hard
    drive is in need of defragmentation, those 11 milliseconds compound very,
    very quickly into a much longer wait times for a file to be read or written.
  • Fragmentation – To further expound on
    the issue of fragmentation, it should be noted that
    fragmented files have almost no effect whatsoever on SSDs. There is an
    upward limit where fragmentation can become a factor, but it is almost never
    reached by the vast majority of drives. Conversely, a fragmented hard
    drive can be expected to experience significant performance degradation. This can be
    corrected by defragmenting the drive, but it is a time consuming
    process that often puts the system out of commission
    until it is being completed.
  • Noise and Heat – Simply put, SSDs
    produce none of the former and almost none of the latter. Meanwhile, hard
    drives produce varying amounts of noise as well as generating their own
    heat. This results in the need for additional cooling to prevent failure due to overheating. This can, once again, be attributed to
    their moving parts.
  • Shock Resistance – As mentioned earlier
    in this report, SSDs are more resistant to damage from drops since they have no moving parts to jostle when an impact occurs. On
    the other hand, disk-based drives are extremely prone to drop damage,
    particularly when they are in active operation. The damage usually occurs
    when the read/write head makes unwanted contact with a platter while it is
    in motion. A common result is damage to one or both components,
    either rendering the drive unusable, or, at best, creating a section of the
    drive that can no longer be used and permanently reducing its capacity. This
    is, of course, an issue that primarily affects laptops, tablets, and other
    portable systems, but it can also impact desktops and servers as damage can
    still occur during a move even if the hard drive is not powered on.
  • Physical Size – Hard drives typically
    bottom out at a standard 2.5" wide form factor. While most consumer-grade
    SSDs are also 2.5" and are designed to be used in the same bays, there
    are much smaller options available. This includes the M.2 variety as
    well as PCI card options (both covered below), either of which can be used in
    systems with little or no room to spare for a 2.5" bay.

Where Solid State Drives Fall Short

  • Capacity – Solid state drives are very
    much behind their more established hard disk counterparts in terms of
    capacity,
    both in availability and in cost per GB or storage. Looking purely at capacity, most off-the-shelf solid state
    drives currently top out at about 4TB per drive. This means that any storage
    installation requiring more than 4TB of space will need multiple drives
    set up in what’s known as a RAID array – a series of individual storage
    solutions working as a single, larger storage solutions with the capacity of
    all included drives combined into a single large partition. While this is
    entirely possible, particularly for enterprise customers, it is often
    impractical as hard drives can easily be acquired in sizes up to 10TB+, or
    more than double what the largest SSD is capable of storing. When
    a very large amount of data is being stored, only a select type of
    customer or business should be using SSDs for the entire task. Most
    customers will have an easier time using a hybrid solution like the one
    described above, with SSDs being used only to store data that is being
    accessed on a regular basis.
  • Cost – While the price of solid state
    drives has fallen dramatically since they were first introduced to the
    general public, they remain the most costly of the mainstream mass storage
    options available. This is due to their newness, the complexity of their
    manufacture, and the relatively smaller number of companies producing
    them when compared to traditional hard drives. To offer a price comparison,
    a 4TB SSD can currently sells for a retail price between $500 and $1,000,
    depending on the model, the type of connector it uses, and various other
    characteristics. Conversely, a 10TB hard
    drive may range from $300 to $700, also
    depending on its characteristics. The difference, particularly
    at the high end of both spectrums, is vast and would add up very quickly for
    an enterprise that could need hundreds of either type of drive to suit its
    needs. Again, this means that the use of solid state drives must be
    carefully considered against the additional cost that their performance
    brings.
  • Long-Term Reliability – This is an area
    of solid state drives that is still somewhat murky. Whereas a hard disk
    drive is likely to have a mechanical component that fails long
    before any issues ever arise with its magnetic storage platters, solid state
    drives tend to eventually fail due to the limited number of times each
    physical block of memory storage can be written.16
    This is a known limitation of SSDs and one that most users will never
    encounter because the number of writes a given block of memory can sustain
    is so large that most drives will fade into obsolescence long before
    approaching their failure point. That said, this should be a serious consideration for
    storage installations where drives will be under constant strain, with
    thousands or tens of thousands of write operations per day. The risk of a
    drive failing due to this issue can be mitigated, thankfully. Ways include
    positioning the drive in a RAID array, essentially dividing the "write wear"
    equally among all included drives, using enterprise-class drives
    specifically designed to handle the pressures of a business usage
    environment, and certain software solutions which can alleviate some wear by
    eliminating unnecessary writes.17 With all this said, the average
    failure rate of mechanical components in hard drives, combined with the fact
    that all enterprise customers should be maintaining multiple, multi-site
    backups of their data, results in a real-world operational environment where
    this particular weakness is a minor nuisance at best.

Connector Technology

Figure 3: A
Solid State Drive’s Sata III Connector and Power Connector

Figure 3: A Solid State Drive's Sata III Connector and Power Connector

Source: Crucial

  • SATA – This is the oldest and most
    common connector found on both consumer-grade and, to a lesser degree,
    enterprise grade SSDs. The acronym stands for Serial AT Attachment and
    represents a technology that has been used to connect hard drives and
    optical drives, as well as other internal system components, for many years.
    It is one of the most widely used due to its presence on nearly all modern
    motherboards and logic boards in everything from consumer-centric desktops
    to enterprise-class servers. Despite the age of the connector technology, it
    continues to be more than a match for the speed modern, non-m.2 SSDs are capable of
    producing with the latest version, version 3.0, supporting transfer rates
    of up to 600MB per second or about 50 to 100MB per second faster than most
    mass market SSDs.18

Figure 4: A
Solid State Drive Equipped with a PCI Express Connector

Figure 4: A Solid State Drive Equipped with a PCI Express Connector

Source:
Plextor

  • PCI Express – This technology
    (Peripheral Component Interconnect Express) is best known for connecting
    system components such as video cards, sound cards, and networking cards to
    motherboards. Because it use powers one of the most demanding aspects
    of computing – video rendering – it was designed for extremely high
    throughput, making it ideal to power SSDs so fast that they outstrip what even
    SATA III is capable of. Currently PCI Express 4.0, the latest version, can
    support transfer rates of nearly 32GB per second. The next revision, PCI Express
    5.0, is expected to double that figure.19,20 For a time,
    this was the only option available to enterprises and enthusiast-level
    consumers who wanted the fastest read and write speeds for demanding data
    processing tasks or gaming and video and sound editing. It should
    be noted that this connector type carries both the data and power for
    the drive, removing the need for a separate power cord as is
    required by the above SATA III drive.

Figure 5: A
Solid State Drive Inserted into a Motherboard’s M.2 Port

Figure 5: A Solid State Drive Inserted into a Motherboard's M.2 Port

Source: Tom’s Hardware

  • M.2 – The term M.2 refers to the connector being
    used, not to the actual transmission technology. In reality, an M.2
    connector situated on a motherboard is capable of connecting a SSD to the
    board’s PCI Express, SATA, or even USB hardware, making it both
    versatile but also potentially limiting. If the M.2 port in question is, for
    example, designed as to be part of the board’s SATA components, it will
    limit any drive attached to it to a transfer rate of about 600MB per second.
    However, if it is connected to the PCI Express components, it can provide
    the full 32GB per second or more mentioned above. Because of this, SSDs
    designed for this type of connection range from the slowest to some of the
    fastest available. One may wonder why M.2 is necessary if it is just
    duplicating the functionality of a PCI Express or SATA port. The answer is
    its size. As can be seen above, M.2 connectors can be situated so that the
    installed SSD is parallel to the motherboard, taking up
    almost no usable space and not interfering with airflow or the installation
    of additional components. Like PCI Express, M.2 ports also provide power to
    the inserted cards, eliminating another cord that would need to be run.

When to Use an SSD Instead of a Hard Drive

The number of use cases that storage technologies are being put to is
currently so numerous that a report many times this size would be required to
even scratch the surface. However, there are some simple facts that can be
summarized to help those trying to decide which storage technology
is best for them, whether they are building a multi-million dollar server farm or
a gaming PC for their home.

  • When speed is of the utmost importance – To put it
    bluntly, solid state drives are not always worth the extra cost associated
    with them. In situations where data will be accessed infrequently, or only
    for simple tasks (playing a movie, storing documents, maintaining
    infrequently used databases), the speed achieved by even an average hard
    drive is more than enough. However, there are tasks that require every bit
    of speed a solid state drive can provide, both for enterprise clients and
    for some private consumers. These include video and sound editing, where
    larger files are constantly being written to and read from the drive;
    gaming, where the speed of an SSD can greatly minimize load times for local
    and online games; and "Big Data" processing, such as large-scale medical
    research projects, massive sensor networks, or scientific endeavors, all of
    which can be greatly expedited with the speed offered by an SSD. In these
    cases, and the various other use cases where speed is an issue, an SSD is
    currently the best option.
  • When a smaller primary drive will do – Most
    consumer-level – and even some enterprise level systems – tend to use SSDs
    only as their primary drive or as the drive on which that PC or server’s
    operating system and most frequently used apps are stored. The reason for
    this is twofold: First, it allows the system to be built with a much
    smaller, cheaper solid state drive, since it only needs enough room to
    hold the operating system and most frequently used apps; second, it offers
    all of the speed benefits that can be provided by an SSD while offloading
    the job of storing larger, less active files to a cheaper hard drive. For
    example, a video editor could get by with a mere 500GB SSD by only
    installing Windows, Adobe After Effects, and a few other apps onto it, while
    keeping the projects that he or she is not actively working on stored on a
    separate, multi-terabyte hard drive. This setup is a fraction of the cost of
    purchasing a multi-terabyte SSD but provides essentially all of the same
    benefits. That said, there is a downward limit to just how small an SSD can
    be practically used in such a scenario. With operating system sizes
    approaching 32GB or more, and many apps nearing or exceeding this limit, careful
    thought must be put into calculating how little room a primary drive can
    actually subsist on.
  • When the drive will be exposed to harsh conditions –
    This use case refers to both frequent motion or drops as well as dust,
    heat, cold, and other extreme situations in which the average hard drive
    would not survive. As previously stated, SSDs do not have moving parts that
    could cause internal damage in the event of a drop, nor do they have
    physical openings into which dust or moderate moisture can encroach.
    This makes them a must-have in products and product lines where toughness and
    resilience are a factor. Care should still be taken, as any electronic
    device can be damaged by harsh treatment. However, the level of
    impact or inclement atmosphere that the average SSD can sustain is many
    times greater than anything the average hard drive is capable of withstanding.
  • When money is no object – The fact is, some
    enterprises, users, and projects simply do not have a set budget that they
    need to adhere to. In this case, SSDs consistently offer the best
    performance currently available. Yes, a RAID array using several
    multi-terabyte drives could easily reach into the tens of thousands of
    dollars. However, when that is scaled against, for example, the budget of a
    Hollywood blockbuster, or the IT spend of a Fortune 500 company, it is very
    often worth the price.

These are just a handful of the situations in which an SSD would be the
correct choice over a hard drive. This is by no means an exhaustive
list. Each potential buyer must examine their own situation, needs, and budget
to determine when the balance between cost and performance has shifted toward
the performance end and when it has tilted towards the financial end.

Summary

[return to top of this report]

Solid state drives are the most advanced
storage technology currently on the market, with ever-increasing speeds and
wider and wider applications. However, they remain a technology that is still
relatively early in its existence. Like most new technologies,
they are often more costly and less versatile than more established
counterparts, in this case, hard disk drives. It remains to be seen if SSDs will
eventually replace hard drives entirely or if they themselves will be
usurped by a newer, even faster form of flash-based storage technology. In any case,
flash-based storage has more than proven that it is a worthy, strong competitor
to the hard drives that have been fulfilling the storage needs of most computers
since the 1980s, particularly for those customers who value performance over
other factors.

[return to top of this report]

References

1 Beard, Brian. "SSD Moving into
the Mainstream As PCs Go 100% Solid State." Samsung Semiconductor. May 2010.

2 Aughton, Simon.
"Dell Gets Flash with SSD Option for Laptops." IT Pro. April 2007.

3 Chen, Shu-Ching Jean. "$199 Laptop
Is No Child’s Play." Forbes. June 2007.

4 Patowary, Kaushik. "Interesting Hard
Drive Facts You Probably Didn’t Know." Instant Fundas. February 2010.

5 Ibid.

6 "Macworld 2008 Steve Jobs Apple
Keynote." Apple. January 2008.

7 Shimpi, Anand Lal. "The MacBook Air:
Thoroughly Reviewed." AnandTech. February
2008.

8 "EMC with STEC for Enterprise
Flash Drives." StorageNewsletter.com.
January 2008.

9 "Solaris ZFS Enables Hybrid
Storage Pools: Shatters Economic and Performance Barriers." Oracle/Sun
Microsystems. June 2008.

10 Ridden, Paul. "Seagate Momentus XT Hybrid
Claimed As World’s Fastest Laptop Drive." Gizmag. May
2010.

11 Webster, Sean "Best SSDs 2021: From Budget SATA to
Blazing-Fast NVMe" TomsHardware.
April 2021.

12 "SSD Ranking: The Fastest Solid State
Drives." FastestSSD. January 2020.

13 Kelion, Leo.
"Seagate Cheetah 15k.6 Hard Drive Review." HotHardware. September 2008.

14 Ibid.

15 Ibid.

16 Kerekes, Zsolt. "SSD Myths and Legends – ‘Write
Endurance.’ " StorageSearch.com. April
2016.

17 Ibid.

18 "SATA Revision 3.0." The Serial ATA Organization. May 2009.

19 "PCI Express 4.0 Frequently Asked Questions." PCI-SIG. Retrieved April 2019.

20
"PCI Express 5.0 Frequently Asked Questions." PCI-SIG. Retrieved April 2019.

About the Author

[return to top of this report]

Michael Gariffo is an editor for Faulkner Information Services. He
tracks and writes about enterprise software and the IT services sector, as well
as telecommunications and data networking.

[return to top of this report]