TY - GEN
T1 - WineFS
T2 - 28th ACM Symposium on Operating Systems Principles, SOSP 2021
AU - Kadekodi, Rohan
AU - Kadekodi, Saurabh
AU - Ponnapalli, Soujanya
AU - Shirwadkar, Harshad
AU - Ganger, Gregory R.
AU - Kolli, Aasheesh
AU - Chidambaram, Vijay
N1 - Publisher Copyright:
© 2021 ACM.
PY - 2021/10/26
Y1 - 2021/10/26
N2 - Modern persistent-memory (PM) file systems perform well in benchmark settings, when the file system is freshly created and empty. But after being aged by usage, as will be the normal mode in practice, their memory-mapped performance degrades significantly. This paper shows that the cause is their inability to use 2MB hugepages to map files when aged, having to use 4KB pages instead and suffering many extra page faults and TLB misses as a result. We introduce WineFS, a novel hugepage-aware PM file system that largely eliminates this effect. WineFS combines a new alignment-aware allocator with fragmentation-avoiding approaches to consistency and concurrency to preserve the ability to use hugepages. Experiments show that WineFS resists the effects of aging and outperforms state-of-the-art PM file systems in both aged and un-aged settings. For example, in an aged setup, the LMDB memory-mapped database obtains 2x higher write throughput on WineFS compared to NOVA, and 70% higher throughput compared to ext4-DAX. When reading a memory-mapped persistent radix tree, WineFS results in 56% lower median latency than NOVA.
AB - Modern persistent-memory (PM) file systems perform well in benchmark settings, when the file system is freshly created and empty. But after being aged by usage, as will be the normal mode in practice, their memory-mapped performance degrades significantly. This paper shows that the cause is their inability to use 2MB hugepages to map files when aged, having to use 4KB pages instead and suffering many extra page faults and TLB misses as a result. We introduce WineFS, a novel hugepage-aware PM file system that largely eliminates this effect. WineFS combines a new alignment-aware allocator with fragmentation-avoiding approaches to consistency and concurrency to preserve the ability to use hugepages. Experiments show that WineFS resists the effects of aging and outperforms state-of-the-art PM file systems in both aged and un-aged settings. For example, in an aged setup, the LMDB memory-mapped database obtains 2x higher write throughput on WineFS compared to NOVA, and 70% higher throughput compared to ext4-DAX. When reading a memory-mapped persistent radix tree, WineFS results in 56% lower median latency than NOVA.
UR - http://www.scopus.com/inward/record.url?scp=85119099327&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85119099327&partnerID=8YFLogxK
U2 - 10.1145/3477132.3483567
DO - 10.1145/3477132.3483567
M3 - Conference contribution
AN - SCOPUS:85119099327
T3 - SOSP 2021 - Proceedings of the 28th ACM Symposium on Operating Systems Principles
SP - 804
EP - 818
BT - SOSP 2021 - Proceedings of the 28th ACM Symposium on Operating Systems Principles
PB - Association for Computing Machinery, Inc
Y2 - 26 October 2021 through 29 October 2021
ER -