Subscribe

Get your free subscription to MES magazine today.

MES

From HDD to flash: Making the switch to meet mission-critical reliability requirements

Military Embedded Systems — May 15, 2006

As flash densities continue to double every 12 months in the same silicon footprint, flash disks are becoming more cost-effective, making them ideal replacements for mechanical HDDs. Solid-state flash disks can operate under the harshest environmental conditions, enabling engineers to achieve the reliability levels military embedded systems must provide.

The life expectancy of a flash disk depends in large part on the flash management tool used. M-Systems TrueFFS software, embedded inside its mSSD family of rugged flash disks, incorporates one of the most effective wear-leveling mechanisms to extend the lifespan of the flash media. Based on the effectiveness of TrueFFS, msystems mSSD flash disks guarantee more than 5 million write/erase cycles, compared with 100,000-300,000 write/erase cycles guaranteed by other flash vendors.

Storing the Operating System (OS), application, and data on a mechanical Hard Disk Drive (HDD) with an average 2 percent failure rate could be the weak link in your military embedded system. Replacing your HDD with a solid-state flash disk managed by effective flash management software provides the reliability and endurance your mission-critical application needs.

HDDs are an inexpensive and high-capacity data storage media with an average failure rate of 2 percent, even in a controlled and air-conditioned environment. But, take the HDD into an embedded military system and this rate can climb to a double-digit figure. Mechanical failures in HDDs come in a variety of guises, such as read/write head failures and motor problems. Conditions of high shock, vibration, altitude, humidity, and extreme temperature ranges can potentially make the HDD the weakest link in mission-critical systems. Equally disturbing, HDDs leave traces of confidential data after erasure.

Replacing HDDs with solid-state flash disks Flash disks are solid-state devices with no moving parts, able to operate in the harshest environmental conditions defined in MIL-STD-810F: Within -40 °C to +85 °C temperature range, absorbing shock conditions at 1,500 g, and withstanding random vibration of 16 g at 80,000 feet altitude. This makes flash disks the ideal storage solution for military embedded systems.

For flash disks to provide true drop-in replacements for HDDs, however, they must have identical dimensions, the same mounting holes, and the same interfaces. Figure 1 shows the most common flash disk form factors: 2.5-inch (laptop size disk) and 3.5-inch (desktop size disk), available in IDE/ATA and narrow/wide SCSI interfaces.

Figure1
Figure 1: the most common flash disk form factors: 2.5-inch (laptop size disk) and 3.5-inch (desktop size disk), available in IDE/ATA and narrow/wide SCSI interfaces

Table 1 compares the characteristics of HDDs with solid-state flash disks, demonstrating the reliability and ruggedness of flash disks.

Table1
Table 1
(click graphic to zoom by 1.9x)

Extending flash life expectancy Solid-state flash disks are a highly efficient nonvolatile memory technology. For mass data storage, NAND flash technology is used instead of NOR flash technology. NAND’s high density and capacity (1 GB memory chip) and lower price per MB make it more attractive than NOR as a mass memory solution. However, some obstacles need to be overcome when using any flash technology as the storage media. One such obstacle is a result of the fact that all flash media has a limited number of write/erase cycles. The life of the device depends on the number of erase cycles guaranteed by the flash vendor, and the frequency with which each specific group of sectors is updated. Because of this limitation, many engineers are concerned about using a flash disk in write-intensive applications.

But, this limitation is one of the most misunderstood parameters of flash. With proper flash management, flash life expectancy can be extended to the point where the write/erase cycle limit is of no consequence, even in the most write-intensive applications. If, on the other hand, flash is not managed properly, its life expectancy can severely diminish the overall life of the flash disk.

In addition to extending the lifespan of flash, flash management handles issues inherent in NAND flash media such as bit-flips and bad blocks, which can compromise data reliability. With the correct error detection and correction mechanism, flash management can guarantee the highest level of data reliability to meet even the strict requirements of mission-critical applications.

The right flash management software There are several methods for managing the flash media when using it to emulate a disk drive. A discussion of two of these methods follows.

Simple flash algorithms map a logical sector to a fixed physical location on the flash. This method quickly causes the flash to wear out when an application updates the same sectors over and over again. Updating the same group of sectors is a very common scenario. All file systems need to maintain some data that describes the allocation of sectors to files. This data is located in a specified area of a disk drive. For example, a File Allocation Table (FAT) file system, which updates the FAT every time a file is extended or concatenated, resides in sequential sectors located at the beginning of the media. If these sectors are updated repeatedly, failures could result after only several thousands of file operations.

Another method uses a sophisticated algorithm that can map the same logical sector to different physical locations. This method, called wear-leveling, ensures that all write/erase cycles are evenly spread across the entire flash array. Static wear-leveling is applied on static files characterized by sectors of data that remain unchanged for very long periods of time. Dynamic wear-leveling is applied on newly written data based on a statistical allocation of physical locations on the flash. The combination of both static and dynamic wear-leveling has the greatest effect on increasing flash life expectancy. Figures 2, 3, and 4 show how wear-leveling extends flash life expectancy.

Figure2
Figure 2: how wear-leveling extends flash life expectancy
(click graphic to zoom by 1.4x)
Figure3
Figure 3
(click graphic to zoom by 1.4x)
Figure4
Figure 4
(click graphic to zoom by 1.5x)

More than extending flash life expectancy, wear-leveling also delays the onset of certain failure mechanisms in the flash. These failure mechanisms can cause entire erase units to become inoperable. When wear-leveling is used, the erase cycle limit of the flash is increased beyond the minimum specified by flash vendors.

msystems’ TrueFFS (True Flash File System) flash management tool, embedded inside its Solid State Disk (mSSD) family of rugged flash disks, incorporates one of the most effective wear-leveling mechanisms in the industry. It supports a reliability level based on the 5-nines concept (99.999 percent).

Life expectancy calculation A number of methods can be used to calculate the life expectancy of the product in which the flash is embedded, depending on the type of flash management used.

The most common flash management algorithm used is called Erase before Write. Every time a sector (512 bytes) is written, a block of sectors (16-128 kB) must be erased. The number of sectors in a block depends on the flash chip used. When using this method, flash disk endurance is calculated based on the number of sectors within an erasable block, the media size (disk capacity), the write/erase cycles guaranteed by the flash disk vendor (typically 100,000-300,000 cycles), and the rate at which data is written.

When using TrueFFS technology as the flash management software, the worst-case flash disk life expectancy is calculated based on the media size, the write/erase cycles guaranteed by msystems (committed to 5 million cycles), the TrueFFS overhead factor (0.5 percent of the disk capacity is used for TrueFFS internal overhead), and the rate at which data is written.

The lifespan calculations in Figure 3 show an example of life expectancy calculations based on Erase before Write and TrueFFS.

Lifespan Calculations

The calculations below compare the “Erase before Write” and TrueFFS technology on a 512MB flash disk with an application that fills the entire media 240 times a day (10 times an hour).

Equation1
Equation 1: Erase before Write” worst case scenario.  Each update to a sector (512bytes) requires an erase of an entire erasable block (32 sectors). This scenario is typical for updating the File Allocation Table (FAT).
(click graphic to zoom by 1.4x)
Equation2
Equation 2: “Erase before Write” best case scenario. Disk updates are done once per full block (32 sectors). An entire erasable block needs to be erased for each  update. This scenario is typical of data acquisition systems that update large files.
(click graphic to zoom by 1.3x)
Equation3
Equation 3: TrueFFS technology best and worst case scenario
(click graphic to zoom by 1.3x)

Summary As flash densities continue to double every 12 months in the same silicon footprint, flash disks are becoming more cost-effective, making them ideal replacements for mechanical HDDs. Solid-state flash disks can operate under the harshest environmental conditions, enabling engineers to achieve the reliability levels military embedded systems must provide.

The life expectancy of a flash disk depends in large part on the flash management tool used. M-Systems TrueFFS software, embedded inside its mSSD family of rugged flash disks, incorporates one of the most effective wear-leveling mechanisms to extend the lifespan of the flash media. Based on the effectiveness of TrueFFS, msystems mSSD flash disks guarantee more than 5 million write/erase cycles, compared with 100,000-300,000 write/erase cycles guaranteed by other flash vendors.