Along with our release of the new M600 SSD for the client computing market, we are very excited to introduce a new feature called dynamic write acceleration. Dynamic write acceleration is the most efficient method of managing the SSD’s NAND media; it optimizes performance while delivering maximum value to the end user. In this post, I’ll provide a little background on what it does and why it’s needed.
Historically, client SSDs have been manufactured with MLC NAND, a cost-effective solution for high-performance data storage. While very space-efficient, modern high-density process nodes can lead to somewhat lower performance, especially on smaller SSD capacities that don’t have very many individual NAND components (and thus less parallelism for programming tasks).
A solution to this problem is to put a cache of dedicated high-speed SLC NAND in front of the MLC main storage. (This works equally well with TLC as the main storage, which is very efficient in space usage, but tends to be a bit slower than MLC.) This method performs well in some cases, but it has some weaknesses too. Most evidently, performance problems show up when the SLC cache is full, and the SSD must write new data in the MLC area. This slows performance significantly, and can exhibit itself especially as the SSD becomes full.
Micron’s engineers have come up with a unique solution in the form of dynamic write acceleration. When this technology is enabled, we can change any given component in an SSD’s NAND array from MLC mode to SLC mode, and back again, at will. Therefore, Micron’s proprietary SSD firmware can accelerate write performance by dynamically adjusting where and how new writes are performed, with no management or manipulation required of the host computer or the end user. In essence, the cache of SLC NAND can change in size and location on the SSD, depending on the nature of the data stream coming to the SSD and on the current fill state of the SSD. This allows our M600 SSD to exhibit maximum performance under a wider variety of workloads—and even when the drive gets closer to being full of saved data.
The following chart shows the amount of NAND that can be dedicated to SLC-accelerated writes. As you can see, the cache is much larger than a competing static cache, even when the drive is mostly full:
Nominal Acceleration Capacity Compared to Static Cache Size of a Competing Technology
Because of dynamic write acceleration, Micron’s M600 SSD exhibits maximum write performance for longer bursts of new data from the host computer than traditional static cache can—even at 50% full. And even at 90% full, the M600 has 2X the SLC caching capacity of competing static caching technology, which doubles the SSD’s ability to handle large bursts of new data from the host computer. The M600 serves up high performance without compromising high-density data storage; it’s a winning storage solution for all client computing applications, especially today’s ultrathin and ultra-light mobile computing applications.
A much more in-depth discussion is available in our Optimized Client Computing With Dynamic Write Acceleration technical brief.