Memory-Driven Design Methodologies For Solid State Drives (SSDs)

The unparalleled cost and form factor advantages of NAND flash memory has driven 35mm photographic film, floppy disks and one inch hard drives to extinction. NAND Flash memory is now expanding its reach in the form of Solid State Drives (SSDs). The basic architecture of SSDs is discussed in Chapter 1. SSDs’ performance and reliability are strictly linked to that of the NAND Flash memories. For this reason, the SSD design flow must follow a bottom up approach that, starting from an accurate knowledge of the reliability of NAND Flash memories, selects the most appropriate error correction strategy to extend the SSD’s lifetime. Chapter 2 will discuss this bottom up approach To fuel the transition from HHD to SSD, NAND must remain very aggressive in terms of ost per bit. When approaching 10 nm technologies, planar NAND is running out of steam. 3D integration turned out to be the most promising alternative. Chapter 3 is about 3D NAND Flash memories. The advent of 3D NAND has introduced significant issues in terms of characterization and system level optimization. In Chapter 4 we’ll show how machine learning algorithms can help designers to optimize the memory reliability. NAND Flash memories are complex systems. Many efforts in the reliability community are devoted to investigating the reliability loss of this storage medium from a cell’s physics point of view. In Chapter 5 we present a different reliability threat related to the high voltage circuitry of the memory: the dependence from the power supply The read disturb is another important problem related to TLC NAND Flash memories since their usage model is predominantly based on read intensive applications. The state of the art qualification methods of Flash memories are performed by uniformly stressing the memory blocks with the same amount of reads. However, by analyzing several workloads, it appears that the read operations can also be concentrated in a specific address range. In Chapter 6, we’ll show the different behavior of a mid 1X TLC NAND Flash under uniform and concentrated read disturb Flash technology in not the only possible medium for SSDs. RRAM is perceived as a reliable alternative to NAND Flash in SSDs for low latency applications. These emerging memories are non volatile as NAND Flash, but with a lower read/write latency and a higher reliability. However, the relatively small storage capacity of RRAM memories integrated so far has limited their usage to specific applications such as saving critical data during power loss events. In Chapter 7 a design space exploration of a 512 GB All RRAM SSD architecture is performed by using a custom developed simulator PCIe DRAM/Flash based NVRAM (Non Volatile RAM) cards are gaining traction in the market because they can be used either as a very fast and secure synchronous write buffer, or to store both critical system data and user data in case of Power Failure. In a nutshell, the host sees the NVRAM card as a bunch of DRAM devices connected over a PCIe bus. If the power suddenly disappears, the on board controller copies the DRAM content to a bank of Flash memories; during this copy operation, a super capacitor supplies the necessary energy. MRAM memories are now mature enough to offer a technically viable alternative to the combination of DRAM and Flash. In Chapter 8 we present a analysis of IOPS and latency (QoS) for both DRAM/Flash based and All MRAM NVRAM cards. Ensuring data protection in Solid State Drives is vital in enterprise application . However, as the reliability of their storage medium is decreasing at the same pace of the technology scaling, this activity is becoming non trivial. In Chapter 9 we model the endurance reliability of an advanced data protection methodology like the intra disk Redundant Array of Independent Disks (RAID) applied to mid 1X Triple Level Cell NAND Flash based SSD. The performed investigations include a parametric analysis of the Uncorrectable Bit Error Rate