Solid state disks (SSDs) were originally designed in the 1980s for
use as cache in real-time performance hungry applications, as well
as mass storage in industrial and military systems, where immunity
against shock, vibration, and extreme temperatures was required.
Throughout the 1980s and 1990s, SSDs remained a relatively expensive
niche product - a high-cost technology only justified for mission-critical
applications such as those found in the avionics and defense industries,
with few and far-in-between deployments in the IT arena.
That was then. Today we are witnessing tremendous improvements
in processing technology due to increasingly complex and demanding
applications. Considering the steady declines in the cost of memory,
demand for SSDs in all applications has significantly increased,
with most of the upward movement and new demand emerging in the
commercial enterprise market.
The two main types of SSDs are DRAM-SSDs and Flash-SSDs. In the
past two decades, IT managers would primarily refer to DRAM-SSDs
when discussing ways to improve I/O and access time performance
in various applications. However, aside from the fast sustained
read/write and low latency characteristics everyone enjoyed, DRAM-SSDs
also had major inherent weaknesses: these storage devices are volatile
and will lose all their data unless an alternative power source
is provided within 10 milliseconds. Moreover, they consume huge
amounts of power and generate excessive heat.
The built-in disadvantage of volatility was partially dealt with
by building, as part of the DRAM SSD, an expensive back-up support
system comprised of hard disk drives (HDDs) and batteries. However,
volatility concerns still remained valid for most applications requiring
high availability. At the same time, as system front end caches
and DRAM SSD requirements continue to rise, two critical challenges
emerge. These include high power consumption and its corollary effect,
high heat dissipation. These issues necessitated higher capacity
and more expensive power supplies coupled with better cooling.
Industry analysts have been relentlessly searching for solutions
to enhance the capability of SSDs to provide higher Inputs/Outputs
per second (IOPS) at lower latencies, while eliminating volatility
and addressing power consumption/heat dissipation concerns.
On the other side of the spectrum are Flash-SSDs. These storage
devices are non-volatile and can retain data for up to 10 years
without system power. At the heart of Flash-SSDs are Flash memory
chips, which consume a fraction of the power DRAM memory draws.
This also amounts to less heat being generated and the possibility,
in some cases, of completely eliminating fans - the last moving
part in a storage system.
What does one obtain? The holy grail of solid-state storage: no
moving parts. "OK. Then why didn't the IT world rush in and
replace their DRAM system caches and SSDs with Flash-SSDs?"
might be a self-evident question. The fact of the matter is that,
until recently, both memory types have been more expensive than
what a wide array of applications could justify, while a single
DRAM chip has been and still is faster than one discrete Flash chip.
How does it all add up?
In the following sections, we will discuss the role of SSDs in
different market segments, the applications that stand to benefit
most from this technology, and price/performance trends in Flash-SSDs.
Page: 1 |
2 | 3 | 4
| NEXT
|
