Abstract

Error-correction codes (ECC) and redundant array of independent disks (RAID) are commonly used to protect NAND-based solid state drives (SSD) from serious noise/disturb. In order to tolerate more errors, common solutions are more complex ECCs and RAIDs. Meanwhile, due to the error-correction mechanisms and flash translation layer, NAND-based SSDs inevitably generate abundant redundant information. The theme of this paper is to show that instead of designing new error correction mechanisms from scratch, the NAND-based SSD reliability can be greatly improved by utilizing current decoders to explore the inherent information between redundant pages. We first propose a joint decoder to recover multiple pages by utilizing redundant pages within a stripe of ECC/RAID, and then we consider a joint decoder to recover multiple pages by utilizing write-amplified (WA) pages. The simulation results show that with the proposed algorithm, a performance gain of as much as 1.8 dB can be obtained. The theoretical study reveals that the joint decoding model can be studied from the perspective of relay channel model, and this helps to explore how much information can be reliably stored with the RAID/WA system in SSD.

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