ScaleXFS - Getting Scalability of XFS Back on the Ring

This course aims to address the scalability issue of XFS journaling in manycore systems by identifying primary causes and proposing key techniques. The learning outcomes include understanding the contention between in-memory and on-disk logging, learning about techniques like Double Committed Item List and Per-core In-memory Logging, and evaluating the performance improvements in ScaleXFS compared to stock XFS. The course teaches skills such as mitigating contention in logging operations and improving global journaling state updates. The teaching method involves presenting research findings, proposed techniques, and performance evaluations. The intended audience for this course includes researchers, developers, and professionals interested in filesystem scalability and performance optimization in manycore systems.