The flash translation layer (FTL) of a modern solid-state drive (SSD) employs the address translation because a host system and a flash memory have different granularity units. For this reason, a large piece of data is broken into several pages, which contain partial-pages. Writing these partial-pages induces lots of read-modify-write operations that increase write amplification. If there is a locality among consecutive write accesses, delaying the cache eviction of partial pages can help reduce the number of flash writes. However, conventional cache replacement policies, which are not based on access characteristics, are not easy to take advantage of the delayed cache eviction. Therefore, in this paper, we propose the access characteristic-based cache replacement policy to reduce the number of cache evictions due to partial-page updates. First, the policy distinguishes eviction priority by data size because the access frequency of cache lines varies depending on the requested data size. Second, the policy manages eviction priority by receiving the data access characteristic, such as reusability, from the host. The proposed policy reduces the number of flash writes due to cache eviction by 10.6 % and 9.4 % on average, respectively, compared to FIFO and LRU policies.
|Title of host publication||Proceedings - 2019 International Conference on Computer Vision Workshop, ICCVW 2019|
|Publisher||Institute of Electrical and Electronics Engineers Inc.|
|Publication status||Published - 2019 Oct|
|Event||17th IEEE/CVF International Conference on Computer Vision Workshop, ICCVW 2019 - Seoul, Korea, Republic of|
Duration: 2019 Oct 27 → 2019 Oct 28
|Name||Proceedings - 2019 International Conference on Computer Vision Workshop, ICCVW 2019|
|Conference||17th IEEE/CVF International Conference on Computer Vision Workshop, ICCVW 2019|
|Country/Territory||Korea, Republic of|
|Period||19/10/27 → 19/10/28|
Bibliographical notePublisher Copyright:
© 2019 IEEE.
All Science Journal Classification (ASJC) codes
- Computer Science Applications
- Computer Vision and Pattern Recognition