FURI | Spring 2024
Optimizing Storage Caching for Zoned Namespace Solid State Drives
Solid State Drives (SSD)’s are widely used for their performance, but possess serious data durability issues. Without the inability to perform in-place writes, garbage collection (GC) must relocate blocks, incurring additional writes and exacerbating wear. Zone Namespace SSD’s (ZNS), a new type of storage device, exposes a device’s logical block addresses as zones to the host, which in turn takes on the responsibility of garbage collection at zone-level granularity. Further, ZNS SSD caching accelerates data accesses to remote storage (e.g., cloud storage) by using local SSDs to store useful data. This project proposes to use the set of blocks with high reuse to determine firstly, the victim zone for GC, and secondly, the blocks within the victim zone to relocate or evict in pursuit of an optimized tradeoff between performance and wear in the Cache.
Student researcher
Lillian Elizabeth Seebold
Computer systems engineering
Hometown: Queen Creek, Arizona, United States
Graduation date: Spring 2026