The History About The Flash Based Ssds

Published Date: 02 Nov 2017

Introduction

Experts in the industry had been predicting the demise of the mechanical hard disk for a long time, but there had been no change for a long time. Slowly, though, the first truly credible rival to the conventional hard disk �C the solid-state drive �C is catching up at a steep pace.

A solid-state drive (SSD) (also known as a solid-state disk or electronic disk) is a kind of storage device that uses integrated circuit assemblies as memory to store data persistently. Another way to describe this is that SSD is a collection of flash memory modules which are deployed in such a way that it can retain its memory even after the computer is shut off. The SSD is totally different than RAM, because it is made up of flash memory as compared to volatile memory, thus the term 'solid state memory' becomes appropriate.

The feature which distinguishes them from traditional electromechanical magnetic disks such as hard disk drives (HDDs) or floppy disks, is that they have no mechanical components like the spinning disks and movable read/write heads as present in the latter. SSDs are typically less prone to physical shock, they are much quieter, and have lower access times as compared to the primitive electromechanical disks. However, while the price of SSDs has been on a constant decline since 2012, they are still about 7 to 8 times more expensive than HDDs.

As of 2010, most SSDs use NAND-based flash memory, which allows them to retain data without power. There are certain applications that require fast access, but not necessarily data retention after power loss, in that case SSDs may be constructed from random-access memory (RAM). Such devices may use external power sources to maintain data after power loss.

There are also Hybrid drives (also known as solid state hybrid drives (SSHD)) that combine the features of SSDs and HDDs, containing a large hard disk drive and an SSD cache to improve performance. The cache is used to access frequently used data. These devices may offer the same performance levels as SSD.

History and Evolution

It is hard to believe that it has been almost 40 years since the first SSD reached the market in 1976. As the years have gone by, the quest for faster, cheaper, higher capacity SSDs has driven the storage technology to a point where no one would have imagined. The evolution of SSD from a bulky and obscenely expensive server accessory to a small consumer box with hundreds of gigabytes of capacity at an affordable price has been nothing short of an amazing accomplishment.

SSDs using RAM and similar technology

SSDs had origins in the 1950s with two similar technologies: magnetic core memory and card capacitor read-only store (CCROS). These memory units originated during the era of vacuum-tube computers. But they became extinct with the introduction of cheaper drum storage units.

SSDs started to get implemented in semiconductor memory in the 1970's and 1980's. They were used for early supercomputers of IBM, Amdahl and Cray; however, the obscenely high price of the SSDs resulted in them being rarely used. In the late 1970s, an electrically alterable ROM (EAROM) was introduced by General Instruments which operated somewhat like the later NAND-based flash memory. Unfortunately, a longer life was not achievable and a host of companies discarded this technology. In 1976 Dataram started selling a product that had only 2 MB of storage and was 19 inches in width called Bulk Core. It was compatible with the minicomputers of Digital Equipment Corporation (DEC) and Data General (DG). In 1978, Storage Technology Corporation (STC) introduced a cabinet-sized memory unit priced at $400,000 that could boast of a capacity of 45 MB. This was the first ever development RAM based solid-state drive.

After that, in the late 1970's, Magnetic Bubble Memory was introduced. Unlike earlier storage units, it had the ability to store data persistently. But eventually it hit a roadblock in terms of its capacity and it never gained widespread recognition.

Axlon was one of the many companies that introduced SSDs for home computers. In 1983, it introduced the PION Interstellar Drive that was used for many models of personal computers and could hold up to 1 MB of data.

In September 1986, Santa Clara Systems introduced BatRam. It was a 4 MB mass storage system expandable to 20 MB using 4 MB memory modules. It included a rechargeable battery used to persistently save the memory chip contents when the computer was shut off.

In 1987, saw EMC corporation entered into the SSD market. They developed solid-state drives for the mini-computer market and they stopped developing SSDs by 1993.

Software-based RAM disks are still in use as of 2009 because they are a lot faster even than the fastest SSD, though they consume more CPU resources and cost a lot.

Flash-based SSDs

The first flash-based prototype was introduced by an Alabama based PC vendor. It was named as DigiPro. It made use of the NOR flash memory chips that were marketed by Intel.

At the beginning of the 1990's, flash-based drives were still costly and rare. On top of that, they were much slower than the RAM-based ones. The cost of SSDs during this time ranged from $14,000 - $47,000.

In 1995, Israeli-based company M-Systems introduced the modern flash-based solid-state drives. The modern flash-based solid-state drives were able to retain data without the use of batteries or other external power source. They had this advantage over RAM-based SSDs but they were by far slower. Since then, SSDs have been used as HDD replacements by the military and aerospace industries, and other mission-critical applications which require the less mean time between failures(MTBF), only possible with the use of solid-state drives. That is achieved due to their ability to endure extreme shock, vibration and temperature ranges.

2003 saw the rise of the cheap flash SSD. It was introduced by Transcend and it included a 40 or 44 pin PATA connector. With those drives priced as low as $50, they were the first flash SSDs to be used extensively. After that, in 2006, Samsung released one of the first mass market flash SSDs. It was a 2.5 inch 32 GB drive with a PATA interface that was aimed as a replacement for laptop hard drives. And in the following year, SanDisk introduced their own version of the flash-based SSD known as SATA 5000.With these came the introduction of wear-leveling technology that enabled those drives to be able to withstand more re-writes than traditional flash media cards prevalent at that time.

By 2009, the SSDs became so fast that the SATA interface used in hard drives became a hindrance. This resulted in the introduction of DDrive X1 in 2009. It plugged into a PCI express slot and offered 4 GB of DRAM storage and 4 GB of flash memory as backup. Similar technology was used in the making of the Fusion IoDrive Duo. It also made used PCI Express slots and it had a specially designed form of flash memory that allowed speeds of up to 1.5 GB/s. It was priced a walloping $5950 at the time of its launch in 2009.

At Cebit 2009, OCZ technology introduced a 1 TB flash-based SSD that used a PCI Express ��8 interface. It reached a maximum write speed of 654 MB/s and maximum read speed of 712 MB/s.

In 2012, Virident Systems introduced their second generation flash-based SSD called FlashMAX II. It packed the industry's highest storage capacity into a small form factor: just ? length x ? height so as to fit in any server with cards providing up to 2.2 TB of storage. These cards can deliver read bandwidth up to 2.7 GB/s and 1.5 million IOPS in a single low-profile card.

Enterprise Flash Drives

Enterprise flash drives (EFDs) were introduced to run with applications that required high I/O performance, reliability and energy.

In the final quarter of 2012, Intel introduced its SDD DC S3700 drive, which aimed at providing consistent performance, an area that had previously not been worked upon much but which Intel claimed was important for the enterprise market. In particular, Intel claims that at the present rate the S3700 won't vary its IOPS by more than 10�C15%, and that 99.9% of all 4 KB random IOs are serviced in less than 500��s.

File System for SSD & its impact on the design of file systems

In conventional Hard disk drives, the storage medium does not degrade from read and write operations. Disk delays were dominated by three factors: seek time, rotational latency, and transfer time. During the seek time, an actuator moves the disk heads to the disk cylinder being accessed. Rotational latency allows a specific disk block to spin under the disk head, and transfer time allows data to be accessed as it passes under the disk head.

As a result, the file systems designed for a regular hard disk were optimized to minimize actuator movement & mechanical wear and tear which is non-existent in Flash based SSDs. Solid State Drives have no moving parts to fail mechanically and hence are not affected by any of the above factors. Each block of a flash-based SSD can only be erased (and therefore written) a limited number of times before it fails.

Therefore the design of File systems for use in SSDs focussed on algorithms which efficiently reduces the number of program/erase cycles while ensuring a levelled wearout without compromising the performance of the device.

Performance Factors

Even though the SSDs are compatible with conventional disk file systems, the life expectancy of the SSD is adversely affected and its performance is suboptimal if the following issues are not managed:

�� Write Amplification: Flash memory blocks have to be explicitly erased before they can be written to. The time taken to erase blocks can be significant, thus it is beneficial to erase unused blocks while the device is idle. If data has to be overwritten in a used block, the SSD follows a Read-Modify-Write cycle in place of a write cycle which reduces the performance of the SSD.

�� Wear leveling: Flash memory devices tend to wear out when a single block is repeatedly overwritten; flash file systems should be designed to spread out writes evenly, Thereby ensuring a leveled aging of all the blocks.

�� TRIM: Due to the limited number of program/erase cycles and the non-overwritable nature of the SSD media, the File system uses the TRIM command to inform the SSD of the garbage data so that it can be erased internally by the SSD when not in use.

Additionally the following features of the operating system should be removed while using a SSD as they have a negative impact on the performance and life of the SSD:

Random access: Disk file systems are optimized to avoid disk seeks whenever possible, due to the high cost of seeking. Flash memory devices impose no seek latency.

Defragmentation: Defragmentation is a process that reduces the amount of fragmentation. It does this by physically organizing the contents of the mass storage device used to store files into the smallest number of contiguous regions (fragments). Fragmentation does not affect the performance of SSD and therefore Defragmentation results in unnecessary wastage of the limited program/erase cycles thereby reducing the life of the SSD.

Metadata: Conventional Filesystems may update the metadata for files such as ��Last Modified/Accessed on�� very often whenever the data is accessed or modified. Rewriting a particular block containing this metadata in the SSD several times may result in faster exhaustion the usable program/erase cycles for that particular block which results in uneven wearing of the medium. Hence such metadata which require modification very often are not stored in SSDs.

NAND Flash

NAND Flash Memory is the basic structural unit of the SSD. NAND is written to at the page level (4KB), but erased at the block level (512 pages). Unless told otherwise, SSDs try to retain data as long as possible because to erase a block of NAND usually means erasing a bunch of valid as well as invalid data and then re-writing the valid data again to a new block. Garbage collection is the process by which a block of NAND is cleaned for future writes.

Diagram from http://www.research.ibm.com/labs/zurich/

Although several file systems were designed to be used in NAND based flash storage devices (like memory cards , pen drives) which don't have inbuilt controllers, SSDs have inbuilt controllers which take care of garbage collection and wear leveling. Enhancements at the flash controller has obviated the need to invest significant effort in re-writing a custom flash file system. It has also alleviated the overhead of transitioning from rotating disks to flash-based storage by exporting a ``flash-disk" that performs well even with existing file systems.

Need for Flash File System

Conventional file systems tend to lay out files with great care for spatial locality and make in-place changes to their data structures in order to perform well on optical and magnetic disks, which tend to seek relatively slowly.

Several File systems specific to flash devices have also been proposed. Most of these designs are based on Log-structured File Systems, as a way to compensate for the write latency associated with erasures.

The design of log-structured file systems was based on the hypothesis that spatial locality will no longer be effective because ever-increasing memory sizes on modern computers would lead to I/O becoming write-heavy because reads would be almost always satisfied from memory cache. This hypothesis synced to the requirements of the SSD. A log-structured file system thus treats its storage as a circular log and writes sequentially to the head of the log.

��This has several important side effects:

Write throughput on optical and magnetic disks is improved because they can be batched into large sequential runs and costly seeks are kept to a minimum.

Writes create multiple, chronologically-advancing versions of both file data and meta-data. Some implementations make these old file versions nameable and accessible, a feature sometimes called time-travel or snapshotting.

Recovery from crashes is simpler. Upon its next mount, the file system does not need to walk all its data structures to fix any inconsistencies, but can reconstruct its state from the last consistent point in the log.��(Cited)

Log-structured file systems, however, must reclaim free space from the tail of the log to prevent the file system from becoming full when the head of the log wraps around to meet it. The tail can release space and move forward by skipping over data for which newer versions exist farther ahead in the log. If there are no newer versions, then the data is moved and appended to the head.

To reduce the overhead incurred by this garbage collection, most implementations avoid purely circular logs and divide up their storage into segments. The head of the log simply advances into non-adjacent segments which are already free. If space is needed, the least-full segments are reclaimed first. This decreases the I/O load of the garbage collector, but becomes increasingly ineffective as the file system fills up and nears capacity.

Several File systems like ext4, JFS, Btrfs support the TRIM function which helps in discarding data in SSDs. However running the garbage collector in real-time during the delete operation impacts the performance heavily. Therefore even though the use of these filesystems improve the life time of the SSD, the performance has a negative impact. Hence the ��discard parameter�� which should be enabled while mounting a filesystem is disabled by default and thus requires a manual modification of fstab file.

The F2FS (Flash Friendly File System) developed by Samsung in Dec 2012 for the Linux operating system takes into account the characteristics of NAND flash memory based storage devices such as SSDs. F2FS was added to the Linux kernel 3.8. It follows the log structured file system approach eliminating some of the known issues like snowball effect of wandering trees and high cleaning overhead. F2FS is designed to run along the Flash Translation Layer (FTL) where tasks such as wear-leveling and garbage collection are left to the FTL and the F2FS provides large-scale write gathering so that when lots of blocks need to be written at the same time they are collected into large sequential writes which are much easier for the FTL to handle. Rather than creating a single large write, f2fs actually creates up to six in parallel.

One area of difficulty is that the shape of an F2Fs (such as section and zone size) needs to be tuned to the particular flash device and its FTL but the hardware vendors are notoriously secretive about exactly how their FTL works.

Comparison of SSD with HDD

In the past few years the SSD market has made immense steps. Together with constantly falling prices, the performance has also increased with each generation. Most people today have to decide between a Solid State Drive (SSD) or Hard Disk Drive (HDD) as a storage component on their machine. So is SSD better or HDD? There is no clear answer to this question; each consumer has to evaluate the decision based on factors like performance, cost, etc. The price per gigabyte advantage is with HDD even though the price of SDD has been falling. Certainly, SSD is the way to go if performance & fast boot-up is your main consideration. We��ll now make a comparison of SSD with HDD storage.

What is an SSD?

The abbreviation SSD stands for Solid State Disk and constitutes a drive in which the information is stored in microchips. It is built from non-volatile memory chips. There are two kinds of memory chips which are used currently in SSDs. SLC (Single Level Cell) and MLC (Multi Level Cell) NAND memory. SLC saves 1 bit (0 or 1) per transistor due to which it can be read or written to more quickly. MLC chips save 2 bits (00, 01, 10 or 11) per transistor. Current SLC chips are good for 100,000 WR/Delete cycles and MLC are good for 10,000 WR/Delete cycles. Another key advantage of SSD drives is durability in comparison to a hard drive which consists of various moving parts making them more vulnerable to shock and damage. A hard drive uses a mechanical arm which has a read/write head that moves around and reads information on the storage platter from the right location. This is what makes SSD faster than HDD. HDD simply requires more mechanical movement to get the required information.

The individual memory cells in SSD are organized into pages of 4KB each which is the smallest unit which can be read or written to. After which the 512 KB (128 pages) are again organized in blocks making SSDs more susceptible to problems. Only whole blocks can be deleted. If you want to rewrite a separate 4KB area, the controller will read out whole 512 KB for data changes in the memory by deleting the block and rewriting the shifted data. There by making a long chain of commands for writing a 4KB data.

For avoiding this, there are various techniques. One is SSD first writes to empty cells and at the same time accesses the memory structure. Other way is to ask the operating system to forward the delete command for which a command known as the TRIM command was established. Thereby it becomes clearly slower to delete the data.

A general SSD uses a NAND-based flash memory which is a non-volatile type of memory. It simply means that when you turn off the device and it won��t forget what was stored on it so as to retrieve it later-on. Earlier on there were rumors saying that the stored data on the SSD would wear off with time. But today this is not the case; you can read/write to an SSD and the data storage will be maintained for lifetime.

Like an HDD, an SDD does not have a mechanical arm to read/write data. It in-turn has a controller which is a embedded processor that performs a set of operations related to writing and reading data. The controller is one of the most important factor that determines the speed of the SSD as it makes decisions related to how to cache, store, retrieve, and clean up data. It also performs various other tasks like error correction, encryption, garbage collection, read and write caching, etc.

What is an HDD?

Hard Disk Drive or HDDs were first introduced by IBM in 1956. It uses a spinning platter on which there is a read/write head. The speed of the platter determines the speed of an HDD i.e. the faster it spins the faster an HDD can perform. Typically today��s laptop drives spin at about 5400/7200 RPM (Revolutions per Minute) compared to server based platters which spin at about 15,000 RPM.

An HDD can store large amounts of data at a very low cost giving it a major advantage when it comes to comparing it with an SSD. Due to which these days 1 Terabyte (1,024 Gigabytes) of storage capacity is very usual for laptop hard drive, and it continues to grow with each passing day. When it comes to cost per gigabyte, it is just about $0.075 per GB for an HDD versus $1.00 per GB for an SSD making it more appealing to consumers who want more storage at a lower cost. When viewed from outside there is barely any difference in appearance of HDDs from SSDs.

Now it��s time to compare SSD with HDD to determine which one is best according to individual needs.

FEATURE

SSD (Solid State Drive)

HDD (Hard Disk Drive)

Power Drawn/Battery Life

Less power used, about 2 to 3 watts giving it 30+ minutes battery boost

More power used, about 6 to 7 watts and therefore uses more battery

Cost

About $1.00 per GB when buying a 240GB drive

Only about $0.075 per GB when buying a 4TB model

Capacity

Available in size of up to 2TB but less costly 256GB are more common

Up to 4TB available

Noise

No noise as there are no moving parts

Some clicks and spinning noise can be heard

Vibration

No vibration. Also they can withstand vibrations of up to 2000 Hz.

Due to spinning of the platters it causes some vibration. Thus, makes them susceptible to crashes and damage.

Heat Produced

Very less heat is produced

Heat not produced but will have a calculable amount more heat than an SSD due higher power draw and moving parts

Write Speed

Ranges from 200 MB/s to 500 MB/s

Ranges from 50 to 120 MB/s

File Opening Speed

Up to 30% faster than HDD

Slower than SSD

Cost and Performance Evaluation, competing technologies

"Solid state disks (SSDs) consisting of NAND flash memory are being widely used in laptops, desktops, and even enterprise servers. SSDs have many advantages over hard disk drives (HDDs) in terms of reliability, performance, durability, and power efficiency. Typically, the internal hardware and software organization varies significantly from SSD to SSD and thus each SSD exhibits different parameters which influence the overall performance. In this paper, we propose a methodology which can extract several essential parameters affecting the performance of SSDs. The target parameters of SSDs considered in this paper are (1) the size of read/write unit, (2) the size of erase unit, (3) the type of NAND flash memory used, (4) the size of read buffer, and (5) the size of write buffer. Obtaining these parameters will allow us to understand the internal architecture of the target SSD better and to get the most performance out of SSD by performing SSD-specific optimizations."(cited)

Increasing speed of procedure and access to critical requests is the momentum for investing in solid state technology. Because after the hardware decisions are created you have to address the multimedia questions. Two storage virtualization technologies noted for their skill to create the SSD presentation spike are slender provisioning and automated tiered storage. You ought to additionally retain storage resource association (SRM) multimedia to automatically trail and report how far capacity is being utilized across tiers, and by the SSDs themselves. The SRM multimedia ought to furnish granular plenty detail concerning use to seize the guesswork out of SSD capacity planning. Are those causes that people are not aware of? Countless resolutions need investment in whole "bricks" and enclosures, that considerably increases the investment. Others permit you to buy SSDs in tinier increments as the data set grows. Additionally, will you be compelled to predetermine volumes and requests for SSD technology? Can you use slender provisioning alongside the SSDs, or are you wasting capacity just to allocate? Slender provisioning way space is merely consumed on these luxurious propels after data is composed, departing as far space free as probable and the propels working at top performance. As slender provisioning is obtaining popularity, insufficient vendors proposal the knowledge for SSDs. One more thought is the skill to automate storage tiering. Incorporating SSDs can enhance presentation, but lacking the skill to vibrantly move data to lower storage tiers, new data stays static on the high-performance drives. This swiftly negates the anticipated benefits..

The solid-state drive (SSD) has achieved great success, not only to make their way in the desktop computer, and mission-critical servers. The solid-state hard drive has been proved to be a breakthrough IO performance and leave random IO performance lags far behind. The random IO is the most database administrators will be concerned about, because this is 90% of the IO mode, as visible on the MySQL database server. I've found that the Intel 520 series and the Intel 910 series are very popular, and they give a good digital random IOPS. However, it's not just the performance, you should be concerned about the problem of failure prediction and health measurement instrument is very important, because the loss of data, is a big NO-NO. However, there is a lot of misunderstanding endurance level SSD, because it is mainly measurements compared to when the rotating disk and even endurance levels, there is a big difference between SSD and HDD, and has a direct impact on the level of SSD endurance.

The smallest unit of a web page can be read from or written to the SSD storage, which is usually the size of 4KB or 8KB. These pages usually organized into blocks between 256KB or 1MB in size. SSD has no mechanical parts, no head or what, they did not seek needs to traditional rotating disk. The reading involved SSD read page, it writes, however, is more problematic. Once you write a page of the SSD, you can simply cover (if you want to write new data), it is the same way with HDD. Instead, you must delete the content, and then write a letter. However, SSD can only do erase a block level rather than at the page level. The meaning of this sentence, the SSD must be re-positioned to be deleted, deleted, and writes the new data block before any valid data block. Broadly speaking, writing average erase + write. Today, SSD controller is intelligent and DO erasing background, such a delay of the write operation will not be affected. These background erase, usually in a process known garbage collection. As you can imagine, if not done in the background erase, then too slow.

These blocks can be gave on the number of blocks in the SSD, so existence is truly eases and write times. Attendance is eases / write series measures. Typically, enterprise-grade MLC SSDs existence of concerning 30,000 remove / comprise cycles, and consumer-grade MLC SSD existence 5000-10000 remove / comprise cycles. This fact clearly displays that an SSD's existence depends on how far period it was written. If you have a write-intensive workloads, next engineer ought to contemplate SSD flounder extra swiftly, contrasted to a read-only workload. This is by design.

Using these two technologies in order to create up for the including of this deeds, and cut the existence of the SSD, builders, burden balancing and over-provisioning. Wear formation works to safeguard the SSD remove and comprise in all blocks to a uniform allocation, that creates a little of the blocks will not perish, and next swiftly supplementary blocks. Over-provisioning of capacity of the SSD is one more method, rise SSD stamina. This is the eases and write alongside the method of period (large capacity SSD) by a colossal populace of block allocation, and furnish a colossal spare area. Countless SSD models furnish the space, for example, a 80GB SSD 10GB of above configuration space, it truly is the size of 90GB, it is described as a 80GB SSD. As such excessive configured SSD manufacturers, could additionally be across the use of the whole SSDs, tear in such a manner, for example, you have merely concerning 75% to 80% of the SSD and the partition is departing the rest RAW space is not visible to the OS / file system. Therefore, the oversupply as away portion of the disk capacity, that provides increased durability and performance.

As engineer may have noticed after reading the above part of this post, its all the more important to be able to predict when a SSD would fail and to be able to see health related information about the SSD.

Here is an example of the Intel SSD, SSD SMART attributes can be used to predict when, SSD fail. These attributes are:

- Available_Reservd_Space: This property reserve block the remaining number of reports. The value of the property, in 100 starts, which means that the space reserved is 100% available. The threshold value for this property is 10, which means that 10% of the availability, this indicates that the drive is close to the end of its life.

"- Media_Wearout_Indicator: This property report NAND media erase / write cycle number and the value of the property is reduced from 100 to 1, as the average erase cycle count increases from 0 to a maximum nominal period Once the value of this property to 1 , does not reduce the number of, although it is likely to be a significant additional wear on the device. the threshold value of the value of 1, this property should not be considered.

Part of the the smartctl tool (smartmontools package), engineers can easily read the values ??of these properties, and then use it to predict failure.

Then, the above information can be used in different ways, we can alert if it is close to the threshold value, or measuring how fast values ??decrease, and then to estimate the drive when the rate of reduction in the use, it may fail.

RAID hard drive is usually used for data protection through redundancy and higher performance, and they found that they use the SSD as. Can often be seen RAID level 5 or 6, the mixed use of SSD read / write workloads because visible using these with the rotating disk write punishment is not sense about SSD because no disk seeking participation, so read - modify - write cycle usually involves parity RAID levels, will not cause a loss of performance a lot. The other hand, the stripe and mirror and raise a lot of SSD SSD provides better performance compared to HDD array read performance of redundant array.

The data protected? Do parity RAID level to provide the same level of data protection and mirroring SSD, because they all think of this? I was skeptical, because, as I mentioned above SSD endurance depends on how much it has been written a lot. Parity RAID configuration, resulting in a lot of additional write, because of the change of parity, of course they reduce the life of the SSD, the same in the case of mirror, I do not know the circumstances of wear out, it can provide any benefits of the SSD, the SSD mirror configuration have the same age, why do it? Write the same amount because the the two SSDs mirror in the array to be received, so life will be reduced, at the same amount of time.

"SSD technology has been developing rapidly. Most of the performance measurements used on disk drives with rotating media are also used on SSDs. Performance of flash-based SSDs is difficult to benchmark because of the wide range of possible conditions. In a test performed in 2010 by Xssist, using IOmeter, 4 kB random 70% read/30% write, queue depth 4, the IOPS delivered by the Intel X25-E 64 GB G1 started around 10,000 IOPs, and dropped sharply after 8 minutes to 4,000 IOPS, and continued to decrease gradually for the next 42 minutes. IOPS vary between 3,000 to 4,000 from around 50 minutes onwards for the rest of the 8+ hour test run

Write amplification is the major reason for the change in performance of an SSD over time. Designers of enterprise-grade drives try to avoid this performance variation by increasing over provisioning, and by employing wear-leveling algorithms that move data only when the drives are not heavily utilized " (cited)

"In terms of performance, solid state devices promise to be superior technology to mechanical disks. This study investigates performance of several up-to-date high-end consumer and enterprise Flash solid state devices (SSDs) and relates their performance to that of mechanical disks. For the purpose of this evaluation, the IOZone benchmark is run in single-threaded mode with varying request size and access pattern on an ext3 filesystem mounted on these devices. The price of the measured devices is then used to allow for comparison of price per performance. Measurements presented in this study offer an evaluation of cost-effectiveness of a Flash based SSD storage

solution over a range of workloads. In particular, for sequential access pattern the SSDs are up to 10 times faster for reads and up to 5 times faster than the disks. For random reads, the SSDs provide up to 200x performance advantage. For random writes the SSDs provide up to 135x performance advantage. After weighting these numbers against the prices of the tested devices, we can conclude that SSDs are approaching price per performance of magnetic disks for sequential access patterns workloads and are superior technology to magnetic disks for random access patterns."(cited)

For over 30 years, a random access memory technology gap between the memory hierarchy and the mechanical hard disk access time. In addition, the gap has been expanded imbalance due to the improvement of these technologies. To this day, the difference between the access time of the random access memory and mechanical disk spans six orders of magnitude. This can not curb their data set random access memory, and generate performance application sought to disk and a serious impact.

Since the identification of the access gap continued research, trying to bridge this gap. Its goal has been to provide persistent storage technology, high-performance, appropriate interfaces and competitive prices and capacity. Trends in recent years studies have shown that Flash can access the technology gap is a good candidate. Flash-based solid-state drives generate too much excitement, but they are still relatively expensive, its performance may significantly change based on access patterns. Is a new flash-based SSD technology and crazy different functions, but it is already mature enough to have been a leading manufacturer of disk arrays. To better understand the performance of some high-end consumer and enterprise-class flash solid-state drive to compare the performance of several mechanical disk-based the flash SSDs behavior, the study. Use IOzone benchmarks to compare, performed a series of different size and pattern of access micro..

Applications

1) Database & online transaction processes

Database and online transaction processes (OLTPs) require maximizing the volumes of data and user with the minimum delay time to access it. To acquire this number of enterprise SSDs are included in storage architecture. Most demanded data can be stored in SSDs whereas less demanded can be kept in HDDs.

2) E-mail

SSDs are widely used in Email Servers. Since for good Email servers the wait and response times should be small so that it can process Email faster with a fast backup. Therefore, Email servers are ideal application of SSD servers.

3) Virtualization

Virtualization dramatically increases the I/O demands on the corporate network through consolidation of multiple applications onto one system. The high I/O demands in a data center can function and be supported without unnecessary delays when deployed with SSDs. These benefits translate directly to a reduction in boot storms and log-in storms.

4) Cloud Computing

Cloud computing allows multiple user to share applications and data which is hosted over the internet. Therefore cloud computing architecture requires a memory which has fast response times as multiple users are supported by single host. To meet this requirement SSDs are included in the storage architecture to meet the high performance demands.

5) Telecommunications

Today, competition between various Telecommunications companies is so high that each organization wants to provide the best services. The ability to access that data rapidly is affected because of I/O bottlenecks caused by data latency. To minimize this data latency SSD are used and also to solve the problem of slow access to customer data.

7) SSD as cache

The performance of SSD dive enables them to use them as Cache drive. In this the most used data is stored in a SSD which acts as a DRAM and the data which is not used frequently is kept in a magnetic disk.

Advantages

1) Shock Resistance

Since it has no moving parts like in a magnetic disk its shock tolerance capacity is 10 time better than other drives.

2) Vibration Resistance

With the absence of moving parts SSDs have more vibration resistance than the other hard drives. Vibration resistance of SSDs is 20x more than hard drives.

3) Improved Response time

With SSDs response time is improved by 75% whereas there is a significant decrease in the daily backup time. If a normal hard drive takes 6 hours for daily back up than in case of SSDs it will take less than 2 hour. This quick response time and decreased backup time leads to customer satisfaction.

4) Less Power Usage

At peak load SSDs uses significantly less power than hard drives which puts them ahead in terms of power consumption. This energy efficiency increases the battery life and gives a cooler computing environment.

5) Higher Durability

Solid State Drives involves non-mechanical design of NAND flash mounted on circuit boards which makes it more durable.

6) Portable

SSDS have good shock and vibration resistance which makes them easy to portable.

7) Faster File Transfer

File transfer becomes really faster with SSDs. Their less access time and response time make file transfer faster. So you don��t have to wait long to transfer larger GB file.

8) Light weighted

SSDs are marginally light weighted than the hard drives. Generally SSDs weight around 70g which is exceptionally lightweight when compared with other storage media like hard drives.

9) Cuts backup time by Half

Its always ideal not to interrupt your workflow during routine maintenance tasks. Using SSDs cuts this routine maintenance to almost 50%.

10) Master Multitasking

SSD can smoothly handle multiple programs so with SSDs you can perform different work in less time and with more efficiency.

Disadvantages

1) Cost

Besides all the advantages and useful application what makes customer really think before buying SSDs is the Cost of SSDs.

2) Lifetime

SSDs have limited lifetime typically between 100,000 and 1 million write/erase cycles. Since writing and deleting is faster this is possible that one may hit this limit soon than

expected.

3) SSD Cannot be used in enterprise application

Although SSDs have high performance but it cannot be used in enterprise application because for enterprise application we need even higher performance end devices like DRAM.

4) Need secondary memory in support to SSD

We cannot use SSDs alone. It is always used with secondary memory which is hard drives.

5) Data Recovery

SSD uses much more nonlinear and complex way of storing data as compared to hard disks therefore it becomes complex to recover data.

6) Software planning consideration

There need a special planning for using SSDs because Sometimes programs have a set of hardware that is certified and then you will need to confirm that is your SSD is supported or not.

7) Calculation of space

File systems have superblocks and inodes whose size is not easy to calculate. Allocating a random space for these file systems would be much costlier in case of SSD.

8) Wear Leveling

Wear leveling is a serious issue with the SSD in which after continuously writing and erasing from SSDs its service lifetime decreases.