Abstract

AbstractIn the recent years, flash storage devices such as solid-state drives (SSDs) and flash cards have become a popular choice for the replacement of hard disk drives, especially in the applications of mobile computing devices and consumer electronics. However, the physical constraints of flash memory pose a lifetime limitation on these storage devices. New technologies for ultra-high density flash memory such as multilevel-cell (MLC) flash further degrade flash endurance and worsen this lifetime concern. As a result, flash storage devices may experience a unexpectedly short lifespan, especially when accessing these devices with high frequencies. In order to enhance the endurance of flash storage device, various wear leveling algorithms are proposed to evenly erase blocks of the flash memory so as to prevent wearing out any block excessively. In this chapter, various existing wear leveling algorithms are investigated to point out their design issues and potential problems. Based on this investigation, two efficient wear leveling algorithms (i.e., the evenness-aware algorithm and dual-pool algorithm) are presented to solve the problems of the existing algorithms with the considerations of the limited computing power and memory space in flash storage devices. The evenness-aware algorithm maintains a bit array to keep track of the distribution of block erases to prevent any cold data from staying in any block for a long period of time. The dual-pool algorithm maintains one hot pool and one cold pool to maintain the blocks that store hot data and cold data, respectively, and the excessively erased blocks in the hot pool are exchanged with the rarely erased blocks in the cold pool to prevent any block from being erased excessively. In this chapter, a series of explanations and analyses shows that these two wear leveling algorithms could evenly distribute block erases to the whole flash memory to enhance the endurance of flash memory.KeywordsFlash MemoryGarbage CollectionLogical BlockCold PoolGarbage CollectorThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call