Linux Filesystem Performance Benchmarks Revealed
New tests on Linux 6.17 highlight performance variations among EXT4, XFS, and OpenZFS for server workloads.
Updated on Sept. 19, 2026 in Data Centers
Live Poll
Do you prioritize data integrity features over raw speed when choosing how to store your files?
Recent Phoronix benchmarks on Ubuntu 25.10 show EXT4 maintaining a narrow lead in overall performance across production server workloads. These tests evaluated various filesystems using a Crucial T705 Gen5 NVMe SSD to measure real-world operational consistency.
Why it matters
Storage architects must balance raw speed, feature sets, and reliability when selecting filesystems for data centers. Reproducible data on performance degradation helps professionals avoid significant losses in IOPS efficiency.
Testing on Linux kernel 6.17 utilized a Crucial T705 Gen5 NVMe SSD to benchmark EXT4, XFS, and OpenZFS. EXT4 demonstrated a narrow lead in geometric mean performance, while F2FS matched or exceeded others in specific random read tests.
The players
Phoronix
This technology publication specializes in hardware and Linux performance testing.
Linux
This open-source operating system kernel serves as the foundational software for most modern data center infrastructure.
The details
The benchmark methodology prioritized performance consistency and metadata operations over synthetic peak speeds. OpenZFS showed notable advantages in concurrent SQLite workloads, providing distinct performance profiles for varying data center demands.
Timeline
September 2026: Phoronix published fresh Linux 6.17 performance numbers.
Early September 2026: MLPerf Storage v3.0 release introduced new database tests.
2024: An arXiv paper studied filesystem performance on massive small-file retrieval.
The Tech Race
These benchmarks contribute to the ongoing industry effort to standardize storage metrics, reflecting the standards set by the MLPerf Storage v3.0 release. This shift underscores a move toward quantifying performance consistency over raw peak speeds in high-demand environments.
Server administrators and developers can leverage these findings to optimize storage configurations, potentially recovering 10% of IOPS previously lost to inefficient filesystem choices. These results provide a practical guide for selecting filesystems that best align with specific metadata or read-heavy application requirements.
The takeaway
Selecting the right filesystem requires moving beyond simple synthetic benchmarks to account for workload-specific metadata operations. Architects should monitor ongoing updates, such as planned Bcachefs improvements, to ensure their storage stack remains performant over time.
Further reading
For additional context on infrastructure performance, visit the Data Centers section.
More information
Review the independent filesystem benchmark results for full methodology details.
Source note: This article includes information reported by WebProNews.
Live Poll
Do you prioritize data integrity features over raw speed when choosing how to store your files?
