TSUE: A Two-Stage Data Update Method for an Erasure Coded Cluster File System

TL;DR

TSUE addresses the high update latency and limited lifespan of erasure-coded cluster file systems by introducing a two-stage update mechanism that separates synchronous log appending from asynchronous log recycling. A three-layer log (DataLog, DeltaLog, ParityLog) and a FIFO-based log pool exploit spatio-temporal locality to dramatically reduce random I/O, network traffic, and write amplification, while enabling real-time log recycling. Empirical results on SSD-based clusters show TSUE delivering substantial throughput gains (up to several times faster than state-of-the-art methods) and extending SSD lifespan by reducing overwrites and erase operations. The approach is implemented in a self-developed ECFS and demonstrates robust performance across real cloud traces and HDD/SSD configurations, indicating practical impact for high-performance, durable erasure-coded storage systems.

Abstract

Compared to replication-based storage systems, erasure-coded storage incurs significantly higher overhead during data updates. To address this issue, various parity logging methods have been pro- posed. Nevertheless, due to the long update path and substantial amount of random I/O involved in erasure code update processes, the resulting long latency and low throughput often fail to meet the requirements of high performance applications. To this end, we propose a two-stage data update method called TSUE. TSUE divides the update process into a synchronous stage that records updates in a data log, and an asynchronous stage that recycles the log in real-time. TSUE effectively reduces update latency by transforming random I/O into sequential I/O, and it significantly reduces recycle overhead by utilizing a three-layer log and the spatio-temporal locality of access patterns. In SSDs cluster, TSUE significantly im- proves update performance, achieving improvements of 7.6X under Ali-Cloud trace, 5X under Ten-Cloud trace, while it also extends the SSD's lifespan by up to 13X through reducing the frequencies of reads/writes and of erase operations.

Figures (8)

• Figure 1: The update process of erasure coding and replica. Note that the operation on $P_1$ is also performed on $P_2$ - $P_M$ although it is not illustrated.
• Figure 2: The workflow of TSUE. There is only 2 copies for DataLog in SSD cluster, and 3 copies for DataLog in HDD cluster. Note: Due to the varying performance of different devices, the retention time of log items in memory differs. For SSDs, which have a lifecycle of approximately 10 seconds (Table. \ref{['table2']}), 2 copies are sufficient to ensure data reliability. In contrast, for HDDs, the number of log copies is designed to be 3.
• Figure 3: Log pool structure of TSUE.
• Figure 4: Architecture of ECFS.
• Figure 5: Performance evaluation of TSUE's update throughput with SSDs.