TY - GEN
T1 - SSD Wear Leveling with Optimal Guarantees
AU - Lange, Tomer
AU - Naor, Joseph
AU - Yadgar, Gala
N1 - Publisher Copyright: Copyright © 2024 by SIAM.
PY - 2024
Y1 - 2024
N2 - Flash-based solid-state drives (SSDs) are fundamental building blocks of modern computer systems. However, due to the unique characteristics of flash media, each part of the SSD is limited in the number of times it can be written to. Maximizing the total number of write operations supported by the SSD is a challenging problem which has been addressed by the storage systems community for more than two decades. This optimization requires minimizing the internal garbage collection overheads of the SSD, and distributing the write operations evenly across the entire device. In this paper we address this challenge, considering it in both online and offline settings. In the online setting, we propose a simple randomized algorithm and prove its optimality. Compared to previous solutions, our algorithm both exhibits better performance and is effective for a wider (and more practical) range of SSD parameters. In the offline setting, we prove that the problem is NP-complete, and present an algorithm with near-optimal performance on any input.
AB - Flash-based solid-state drives (SSDs) are fundamental building blocks of modern computer systems. However, due to the unique characteristics of flash media, each part of the SSD is limited in the number of times it can be written to. Maximizing the total number of write operations supported by the SSD is a challenging problem which has been addressed by the storage systems community for more than two decades. This optimization requires minimizing the internal garbage collection overheads of the SSD, and distributing the write operations evenly across the entire device. In this paper we address this challenge, considering it in both online and offline settings. In the online setting, we propose a simple randomized algorithm and prove its optimality. Compared to previous solutions, our algorithm both exhibits better performance and is effective for a wider (and more practical) range of SSD parameters. In the offline setting, we prove that the problem is NP-complete, and present an algorithm with near-optimal performance on any input.
UR - http://www.scopus.com/inward/record.url?scp=85194141043&partnerID=8YFLogxK
M3 - منشور من مؤتمر
T3 - 2024 Symposium on Simplicity in Algorithms, SOSA 2024
SP - 306
EP - 320
BT - 2024 Symposium on Simplicity in Algorithms, SOSA 2024
A2 - Parter, Merav
A2 - Pettie, Seth
T2 - 7th SIAM Symposium on Simplicity in Algorithms, SOSA 2024
Y2 - 8 January 2024 through 10 January 2024
ER -