Because all solid state flash products are not created equal and flash storage is finding its way into more and more embedded computing applications, system designers should understand the critical tradeoffs between competing technologies when evaluating flash products. Most commonly, the endurance and reliability required in end-user applications help dictate the appropriate storage technology to use. Two well-known flash storage technologies, Single Level Cell (SLC) and Multi Level Cell (MLC), offer distinct advantages, depending on a user’s needs.
The Technical Distinction
The difference between MLC and SLC can be found in the voltage level treatment and control within the cells. SLC technology stores a single bit in two binary states in each cell (0 or 1), while MLC stores two bits in four binary states in each cell (00, 01, 10 or 11) (Figure 1). In either case, the binary states are determined by the differentiation in charge levels on the floating gate. Advances in MLC technology are turning out 3- and 4-bit chips with 8 and 16 binary states per cell, respectively.
The downside is that increasing the number of potential binary states within a cell reduces the delta between voltage thresholds, blurring the distinction between cell values, especially in the face of cell degradation over the life of the drive. This drives the need for tighter and more time consuming error correction algorithms and is the reason why SLC flash is faster than MLC.
Wider design and operating margins in SLC flash lead to higher reliability, better endurance and longer life when compared to MLC flash. Drive manufacturers specify endurance as the number of block level write/erase cycles allowed before errors rise to unacceptable levels.
Today those numbers stand at 100,000 cycles for SLC compared to 10,000 or less for MLC (depending on the feature size) – a 10-fold performance delta.
The voltage level of a flash cell or transistor is determined by the buildup of electrons creating a charge on the floating gate of the transistor relative to the substrate. An oxide layer on the floating gate acts as an insulator between the gate and the substrate. Over the course of repeated write/erase cycles, the oxide layer begins to wear thin, allowing electrons to leak back on to the substrate, possibly changing the value of the cell. This effect is exacerbated in MLC devices. In addition, higher temperatures common to military and industrial applications further reduce the ability of the cell to retain the correct voltage level.
Wear leveling and other memory management features boost drive level endurance into the millions of cycles. Memory management aside, many flash users simply can’t tolerate the increased possibility of data errors, so SLC flash manufacturers expect to gain market share in critical applications. Because wear leveling works by ensuring data is distributed equally to all good blocks rather than cycling the same block, it becomes extremely important for the less durable MLC devices. Figure 2 provides endurance comparison data using an 8 GB flash drive and static wear leveling. The data in this example indicate that SLC flash will endure tenfold longer than MLC flash using a range of four typical data rates as examples. In critical storage applications, this can be a crucial distinction.
Data Density & Bandwidth
MLC technology allows more data per unit area to be packed onto a chip compared to SLC, but there is a reliability and performance price to pay for higher densities. Considerations for these tradeoffs are what make SLC flash technology so essential to many mission critical applications.
The need for tighter voltage tolerances required by MLC flash lead to complicated erase and programming processes. The potential for higher error rates in discerning more closely packed voltage thresholds requires the implementation of 4 bit error correction code (ECC) for MLC flash compared to 1 to 2 bit ECC in SLC flash. That additional processing overhead results in MLC write rates of less than a third of SLC and read rates of less than half.
Back To Where It All Began
The original EPROMS of the late 70s and early 80s were precursors to today’s SLC flash storage technology. Intel soon developed a method to boost the charge in EPROM cells allowing more than two detectable charge levels and multi-bit EPROMS were born, leading the way to MLC flash.
Although MLC technology patents disclosed by SanDisk appeared in the early 90s, additional technical fine-tuning and market analysis delayed MLCs introduction by Toshiba and SanDisk until 2000. In those early days, flash with either SLC or MLC were shipping in consumer products and the underlying technology was not necessarily disclosed as long as the end-use requirements for those early applications were satisfied.
Mission critical applications, though, need the reliability of SLC flash. The majority of flash manufactured today is MLC, since capacity per unit area and price are driving factors in the large consumer products market they serve. It is estimated that about 80% of flash products shipped in 2008 were MLC with volumes that drove prices down compared to SLC. Pricing for SLC-based products are about 30% higher than MLC. Market differentiation is now beginning to take place as both technologies mature and SLC-based devices are showing up in industrial and military applications that demand the high reliability and endurance for which SLC technology is known.
Determining the Right Technology
So how does an end user in the embedded market determine which technology to use? Environmental requirements are the first consideration. All flash technology is virtually impervious when it comes to shock and vibration, but this is not the case where operating temperatures are concerned. MLC flash is recommended for use in commercial temperature environments only – for example less than 55°C at the top end. Higher temperatures speed up the cell wear problem mentioned above. The trend in many military applications, as well as some industrial applications, is for rugged equipment to operate at temperatures up to 85°C.
SLC flash is better suited to write intensive applications, although the SLC/MLC decision here is less straight forward. The end user should be very much aware of the read/write ratio for the application. Consider that, in an application with a read/write ratio of 70% to 30%, SLC flash outperforms MLC by a factor of 20 to 1 as measured in hours of data written before the drive begins to fail. Since drive capacity plays into endurance, that same application would require an MLC drive capacity of up to 20 times larger than an SLC drive in order to meet a five year service life. The lower initial investment costs of MLC flash are far outweighed by the longer term maintenance and replacement costs they may represent.
SLC-based flash drives are manufactured in a variety of form factors including the popular Compact Flash (CF) I and II as well as form, fit and function replacements for 1.8” and 2.5” hard drives. Capacities as of this writing are at a respectable 16 GB top end CFI and 128 GB in 2.5” drives and continue to grow rapidly.
Embedded storage products in standard PMC and 6U VME and CompactPCI form factors, such as those offered by Elma Electronic’s Systems Division, use SLC flash drives to meet the needs of applications demanding high reliability and long life. Dual drive 6U network attached storage (NAS) products allow redundant (RAID) capability with two 2.5” SLC flash drives per slot, creating one of the most reliable storage products in the embedded computing industry (Figure 3).
What’s in Store?
SLC and MLC will each continue to find homes in embedded computing environments, since both offer unique cost, reliability and throughput advantages. The key for designers is knowing which solid state flash product fits best in a specific application.