Garbage collection and wear leveling for flash memory: Past and future

Abstract

Recently, storage systems have observed a great leap in performance, reliability, endurance, and cost, due to the advance in nonvolatile memory technologies, such as NAND flash memory. However, although delivering better performance, shock resistance, and energy efficiency than mechanical hard disks, NAND flash memory comes with unique characteristics and operational constraints, and cannot be directly used as an ideal block device. In particular, to address the notorious writeonce property, garbage collection is necessary to clean the outdated data on flash memory. However, garbage collection is very time-consuming and often becomes the performance bottleneck of flash memory. Moreover, because flash memory cells endure very limited writes (as compared to mechanical hard disks) before they are worn out, the wear-leveling design is also indispensable to equalize the use of flash memory space and to prolong the flash memory lifetime. In response, this paper surveys state-of-the-art garbage collection and wear-leveling designs, so as to assist the design of flash memory management in various application scenarios. The future development trends of flash memory, such as the widespread adoption of higher-level flash memory and the emerging of three-dimensional (3D) flash memory architectures, are also discussed.

Error Characterization, Mitigation, and Recovery in Flash-Memory-Based Solid-State Drives

“…By providing this indirection between address spaces, the FTL can remap the logical address to a different physical address (i.e., move the data to a different physical address) without notifying the host. Whenever a page of data is written to by the host or moved for underlying SSD maintenance operations (e.g., garbage collection [35,202]; see below), the old data (i.e., the physical location where the overwritten data resides) is simply marked as invalid in the physical block's metadata, and the new data is written to a page in the flash block that is currently open for writes (see Section 3.4 for more detail on how writes are performed).…”

“…Over time, page invalidations cause fragmentation within a block, where a majority of pages in the block become invalid. The FTL periodically performs garbage collection, which identifies each of the highly fragmented flash blocks and erases the entire block (after migrating any remaining valid pages to a new block, with the goal of fully populating the new block with valid pages) [35,202]. Garbage collection often aims to select the blocks with the least amount of utilization (i.e., the fewest valid pages) first.…”

“…WA (write amplification) is the ratio between the amount of data written into NAND flash memory by the controller over the amount of data written by the host machine. Write amplification occurs because various procedures (e.g., garbage collection [35,202]; and remapping-based refresh, Section 5.3) in the SSD perform additional writes in the background. For example, when garbage collection selects a block to erase, the 6 pages that are remapped to a new block require background writes.…”

RRWL: Round Robin-Based Wear Leveling Using Block Erase Table for Flash Memory

“…To manage this disadvantage, wear leveling techniques have been proposed. Wear leveling increases the lifetime of flash memory by equalizing the number of erases for each block and uses various information tables [1].…”

PBGC: Proxy Block-Based Garbage Collection for Index Structures in NAND Flash Memory

“…However, µ-Tree and µ * -Tree overlook the fact that an additional update propagation problem occurs during garbage collection, whereas they only focused on record operations. D-IPL, AD-Tree, and BbMVBT are possible for performing garbage collection, but their garbage collection has a limited range of blocks to reclaim invalid pages for space in use, and they are difficult to apply to existing wear leveling techniques without additional modifications [7].…”