John Shawger, Arnav Jhingran, Andrea Arpaci-Dusseau, and Remzi Arpaci-Dusseau, University of Wisconsin – Madison
Modern distributed storage systems exhibit cross-layer data duplication, writing data to disk once during a consensus phase and again during a local database logging phase. The result is poor performance and significant write amplification. To remedy this cross-layer redundancy, we design and implement Cross-Layer Log (XLL), a shared log built upon the principle of key-value separation. We use XLL to deduplicate updates within a distributed key-value store (TiKV), leading to a 5.5x increase in write throughput while reducing write amplification by 73%. We also demonstrate the effectiveness of our crash recovery protocol in maintaining data integrity.
NSDI '26 Open Access Sponsored by
King Abdullah University of Science and Technology (KAUST)
Open Access Media
USENIX is committed to Open Access to the research presented at our events. Papers and proceedings are freely available to everyone once the event begins. Any video, audio, and/or slides that are posted after the event are also free and open to everyone. Support USENIX and our commitment to Open Access.
author = {John Shawger and Arnav Jhingran and Andrea Arpaci-Dusseau and Remzi Arpaci-Dusseau},
title = {{XLL}: {Cross-Layer} Logging for Data Deduplication in {Consensus-Based} Storage},
booktitle = {23rd USENIX Symposium on Networked Systems Design and Implementation (NSDI 26)},
year = {2026},
isbn = {978-1-939133-54-0},
address = {Renton, WA},
pages = {2619--2634},
url = {https://www.usenix.org/conference/nsdi26/presentation/shawger},
publisher = {USENIX Association},
month = may
}
