More and more consumer electronics, such as personal computers and smart phones, adopt NAND flash memory as their data storage devices. However, garbage collection (GC) operations in flash memory may degrade both the performance and lifetime for consumer electronics due to the extra page reads and page writes caused by GC. Greedy algorithm, which selects the block with the least valid pages for GC, is usually used for victim block selection so as to reduce page movements. In this paper, to further improve GC performance, we propose a novel caching-aware GC scheme, and its basic idea is to take into account the pages cached in memory so as to further reduce page movements during GC. We also develop a consistent caching scheme to ensure data consistency and revise the victim block selection process with a lightweight caching-aware block selection scheme. We validate the performance of our caching-aware GC on FlashSim, and results show that it improves the performance of SSDs and prolongs their lifetime compared with conventional greedy algorithm.

中文翻译：

---

缓存感知垃圾收集可提高NAND Flash SSD的性能和使用寿命

越来越多的消费电子产品，例如个人计算机和智能电话，都将NAND闪存用作其数据存储设备。但是，由于GC导致额外的页面读取和页面写入，闪存中的垃圾回收（GC）操作可能会降低消费类电子产品的性能和使用寿命。贪婪算法通常选择有效页数最少的块作为GC，通常用于受害者块的选择，以减少页面移动。在本文中，为了进一步提高GC性能，我们提出了一种新颖的可感知缓存的GC方案，其基本思想是考虑缓存在内存中的页面，从而进一步减少GC期间的页面移动。我们还开发了一个一致的缓存方案，以确保数据的一致性，并使用轻量级的缓存感知块选择方案来修改受害者块选择过程。我们验证了我们在FlashSim上具有缓存意识的GC的性能，结果表明，与传统的贪婪算法相比，它可以提高SSD的性能并延长其寿命。