We noticed you're blocking ads.

Keep supporting great journalism by turning off your ad blocker.

Questions about why you are seeing this? Contact us

Font Size

- Aa +

Mon 12 Jan 2009 04:00 AM

Font Size

- Aa +

Storage anatomy

Every desktop PC and notebook has a non-volatile storage device to store data. Here, WINDOWS explains exactly what makes hard drives and newer solid state drives tick.

Every desktop PC and notebook has a non-volatile storage device to store data. Here, WINDOWS explains exactly what makes hard drives and newer solid state drives tick.

Storage devices, whether they are the tried and tested hard disk drive (HDD) or newer solid state drives (SSDs), have long been regarded as rather uninteresting, compared to the other components sitting within a machine.

However, without a storage drive your PC or notebook would be of absolutely no use because the rest of the components would have no data to work with.

Non-volatile storage devices (these do not lose data even when the PC is switched off) not only store the operating system (OS) that you need to use the PC but they also act as a repository for documents, digital images, music files, games and more.

Today, as consumers, we have more choice in terms of what storage device we want to use in our machine than we did two years ago. Rather than just deciding on the capacity, spindle speed and interface the drive uses, we now have a choice of a different drive type (or technology) altogether.

SSD technology is based on high-speed flash memory and although it is new to the consumer market segment, it has actually been present in the workstation and server markets for a lot longer.

The reason this technology remained in these markets for so long and has only now crossed over to the consumer segment is because of its extremely high price. Now though, thanks to advancements in technology, the drives are cheaper to manufacture and now fall into a price bracket that mid- or high-performance desktop or notebook users will agree with.

Before we look at SSDs however, lets take a look at the old workhorse of the computer world first, the standard hard disk drive.

Driving hard since the 50s

Hard disk drives were invented in the 1950s and in their first form they were fairly large and measured 20-inches in diameter (today's desktop-aimed drives are 3.5-inch). One of the earliest disk drives developed by IBM offered 30Mbytes of fixed storage and 30Mbytes of removable storage and thus this drive was named ‘Winchester' after the 30-30 variant of that rifle.

Though this Winchester drive offers an insignificant amount of space compared to today's HDDs, the Winchester name is still used to describe hard drives as the underlying technology is virtually identical.

Of course, although the basic underlying technology has remained the same, the individual components that comprise a hard disk drive have been improved tremendously. They are not only faster, more robust and denser but are also cheaper to manufacture and thus the price of hard disk drives has been steadily dropping over the last few years. Today, you can get a 7200rpm, 1.5Tbyte (1500Gbytes) HDD for well under US $500. The same money 10 years ago would have, most likely, got you a 2.1Gbyte 5400rpm disk.

A hard drive is made up of two distinct pieces both of which will not function without the other being present. The first piece is visible to the naked eye and sits on the underside of a hard drive.

This is a printed circuit board (PCB), which contains the circuitry for the interface connecting the drive to the computer as well as the cache memory, which is used to improve a drive's reading and writing performance. This circuit board also has an ID chip that allows the hard drive to identify itself to the computer's BIOS.

Perhaps most importantly, the PCB is responsible for assembling the magnetic domains on the drive's platter into bytes (known as a read operation) and, on the other hand, can turn data bytes into magnetic domains (known as a write operation). Beyond this, the components that actually perform the physical read/write operations sit on the inside of the drive, out of plain view.

What lies beneath?

The inside of a HDD comprises the components that are actually used to perform reading and writing operations. (Note that a hard drive will fail if it is ever opened regardless of whether it is on or off at the time.)

The components visible are the platters (the shiny discs) and the arm which sits over them. Although the arm looks like it is making physical contact with the platters, it is actually sitting a few millimeters off the surface of the disc. Should these two components ever make contact, it would lead to disk failure. The arm holds the read/write heads that are responsible for transferring data to and from the drive's platters.

It is able to move the heads from the hub of the drive (at the centre) to the outer edge, meaning it is able to cover the entire surface of the platter.(When reading or writing data to the drive's platters, the arm can move as many as 100 times per second in a modern 15,000rpm HDD.)

The platters themselves spin at a certain speed between 5400rpm to 15,000rpm (depending on the drive) and as they are moving, the read/write heads must pick out the spaces on the platters to either read or write data to.

Today, data is generally stored on both sides of the storage platter and thus a separate read/write head is required for each platter surface. So, if you take the case of a hard drive with two platters (four storage surfaces) for example, it would need four read/write heads in all.

A drive's capacity is determined by how dense its storage platters are. This density is generally referred to as ‘areal density'. This is one aspect of the hard drive that has evolved tremendously since IBM introduced the RAMAC hard disk in 1956. This particular drive featured platters that offered a density of two thousand bits per square inch. Today however, drives offer densities greater than 100 gigabits per square inch.

Dense storage platters enable manufacturers to build large hard drives as more data can be packed into the same amount of physical space. At the same time, denser platters also help a drive perform better in terms of reading and writing because the read/write heads don't have to travel as far to perform their operations. How it works

Data is stored on the drive's storage platters in sectors and tracks. On the disc, sectors are pie-shaped wedges that split a disc into several or more pieces whilst the tracks are concentric circles spanning the entire disc.

Each sector on the disc is made up of a fixed number of bytes and, generally, an operating system (OS) such as Windows XP or Vista will group a number of sectors into clusters.

When a HDD is formatted using a ‘low level format', the sectors and tracks are setup and the starting and end points of each sector are written onto the storage platter. Once this type of format is completed, it is then necessary to perform a high-level format so the file allocation table (used by the OS) can be written into the sectors. Once this is done, data can be written to and read from the drive's platters.

When an OS needs to load a file, it will simply refer to the file allocation table, which will in turn look up exactly which track and sectors hold the requested data. The read/write heads will then move across to that area of the storage platter and read the data. When data needs to be written, the OS will use the file allocation table again to determine where there is no data stored. Finally, the read/write head will move over to this area above the platters and begin writing data.

That covers HDD operation; let's move onto SSDs.

Don't do it!

There are a number of things that can cause a hard drive to come to a pre-mature death. Chief among these is excessive movement while the HDD is working. Because the read/write heads hovers just a few millimeters above the magnetic platters that actually store data, sudden movement could cause the head to make contact with the moving platters.

This is known as a head crash and in most cases, this form of HDD failure is impossible to recover from without the use of very expensive equipment. Thus it's very important that you avoid abruptly moving the PC or notebook when it's on or else you could soon have a catastrophic head crash.

You can also make sure your drive lives a little longer by not turning your machine on and off too often. When a machine is turned on, the hard drive's motor must quickly spin its platters from rest to the correct operating speed - measured in revolutions per minute (RPM) - such as 7200rpm or 10,000rpm so the computer's BIOS can detect it properly. This puts serious strain on the motor and frequently powering on the PC or notebook will eventually cause motor failure.

On the SSD train

Solid State Drives (SSDs) are a new entrant to the consumer PC market however, as we said earlier, this type of drive has been available on the performance-intensive workstation and server markets for sometime. SSDs are built to act as normal hard drives so, to an end user, there would be no difference in terms of setup and usage.

A number of firms now offer SSD models including Patriot Memory, Silicon Power (pictured), Super Talent and OCZ. Of the existing standard hard drive manufacturers, only Samsung and Toshiba are currently shipping SSD drives though others are expected to follow over the course of 2009.

The inner workings

Non-volatile NOR and NAND flash memory is used extensively by today's SSDs. Both use an array of memory cells made from floating-gate transistors to function and there are two types of cells; single-level cell (SLC) and multi-level cell (MLC).

The difference between these two comes in the form of the amount of data stored in each ‘cell'. SLCs store 1-bit of data per cell whereas MLCs store 2-bits per cell. Physically, SLC and MLCs both occupy exactly the same amount of die space so a MLC offers double the capacity but at the same price level.

Programming the NOR or NAND flash memory that powers SSDs is a complex process and one that is completely different from the process used by hard drives to read and write data. In a NOR gate flash, each and every cell looks similar to a metal-oxide-semiconductor field-effect transistor (MOSFET). (A MOSFET is a device used to boost or switch electronic signals.) The difference is that the transistor here has two gates rather than one.

To the top of the MOSFET lies a control gate (CG), which sits above a floating gate (FG), both of which are insulated by an oxide layer. The FG is set between the CG and the MOSFET channel. Since the FG is electrically isolated by its insulating layer, any electrons that reside here are normally trapped and won't dissipate for years under normal conditions.

When the FG is charged, it partially cancels the electric field from the CG, which then modifies the voltage threshold (VT) of the cell. During a read-out, a voltage is used on the CG and the MOSFET will then either become a conductor or an insulator depending on the VT of the cell. This is controlled by the charge on the floating gate.

The current flow through the MOSFET is detected and then forms binary code (either a 0 or a 1), which represents the data that is being read from or written to the SSD. In a MLC that stores 2-bits per cell, the amount of current flow is sensed more accurately in order to determine the level of charge on the FG. NOR flash cells by default have a logical value that equals to a binary value of ‘1' and this is because current will flow through the channel when the appropriate voltage is supplied to the CG. To change the value of the NOR flash to binary ‘0', a higher voltage must first be applied to the CG.

This turns the channel on making it possible for electrons to flow from the source to a drain. The source-drain current is then high enough that it causes some high energy electrons to jump through the insulating layer on the floating gate. This process is known as hot-electron injection.

To return a NOR flash cell to its original value of ‘1' (erasing data), a large voltage of the opposite polarity is applied between the CG and source and this pulls the electrons off the FG via a process known as quantum tunneling.

Similarly, NAND flash memory relies on tunnel injection for write operations and tunnel release for erasing data.

With the nitty gritty out of the way, lets examine the pros and cons of SSDs. The good

SSD drives offer a number of advantages over standard hard drives with perhaps the biggest advantage being higher performance. As SSDs don't have to rely on moving read/write heads to perform data operations, they are extremely fast in terms of access time (the time taken to seek out and read data or seek out an empty space to write data).

So whereas a typical 7200rpm hard drive might have an access time of 14.5ms, even the most basic SSD will offer sub 2ms figures. This gives SSDs a distinct performance lead over HDDs in heavy use scenarios where hundreds of thousands of random read/write operations occur.

On the data transfer front, these drives again offer excellent performance because they rely on flash memory rather than rotating storage platters. As a result, SSDs can speed up software load times tremendously.

The performance these drives offer is also far more constant than it is with hard drives and this is because the access time is almost constant, as it doesn't depend on the location of data on a physical disc. As a result, even file fragmentation doesn't negatively impact a SSD's performance.

The lack of moving parts also means that SSDs can start-up and be ready for immediate use far quicker than a standard HDD can. A regular hard drive must spin-up before it can be detected by the PC and OS and this adds a few seconds to a machine's boot time.

SSDs are also quieter than standard HDDs and this is again due to the lack of moving parts. Another advantage that stems from this is that SSDs are far more tolerant in terms of dealing with shocks, high altitude, vibration and extreme temperatures (hot and cold).

SSDs are designed so that the drives themselves know when failure is likely to occur. Getting close to this stage, a SSD will attempt to fail on the ‘next erase' or ‘next write' operation rather than on the ‘next read' operation. With the first two, you won't lose any data as the OS will just report an error saying the drive cannot be written to. With HDDs, failure generally occurs on ‘next read' and this, more often than not, causes data loss.

The bad

Although SSDs offer a number of advantages they are far from perfect and this is why the transition from HDD to SSD will be fairly long drawn out. The biggest issue plaguing SSDs at the moment is high cost. Although they are now much cheaper than before, the cost per gigabyte of a SSD is still extremely high compared to a traditional HDD. For example, a 64Gbyte SSD might offer a cost per gigabyte of US $5 or higher compared to most hard drives which are now sub $1.

Beyond the high cost, SSDs are also available in only limited capacities. The largest SSD on the market offers a capacity of 512Gbytes whereas traditional hard drives now offer capacities as high as 1.5Tbytes at present.

SSDs also offer limited write and erase cycles after which point they cannot be programmed in any way. For MLC based SSDs, you can expect between 1000 and 10,000 write cycles whereas SLCs are closer to 100,000.

Lastly, although SSDs offer better transfer performance than standard HDDs in terms of read operations, SSDs can be quite slow when it comes to writing. This is due to the fact that SSDs rely on fairly large erase blocks, which makes them slower than regular hard drives when writing small chunks of data.

One of the few

Samsung is one of the few standard HDD manufacturers to offer a wide-array of SSD drives. Vendors Hitachi, Seagate and Western Digital are expected to enter the SSD market in the coming months and years.

Arabian Business: why we're going behind a paywall

For all the latest tech news from the UAE and Gulf countries, follow us on Twitter and Linkedin, like us on Facebook and subscribe to our YouTube page, which is updated daily.