On October 24th, the 29th ACM Symposium on Operating Systems Principles (SOSP) was held in Koblenz, Germany. The research paper on Whole-system-persistent Microkernel Operating System by the Institute of Parallel and Distributed Systems (IPADS) at School of Software, the School of Electronic Information and Electrical Engineering , Shanghai Jiao Tong University, was awarded the Best Paper Award. This marks the first time since the inaugural SOSP in 1967 that a research paper completely done by Asian researchers has received the Best Paper Award at the flagship conference of operating systems (i.e., SOSP/OSDI).

Paper introduction

Whole-system persistence promises simplified application deployment and near-instantaneous recovery. This can be implemented using single-level store (SLS) through periodic checkpointing of ephemeral state to persistent devices. However, traditional SLSs suffer from two main issues on checkpointing efficiency and external synchrony, which are critical for low-latency services with persistence need.

The decentralized state of microkernel-based systems can be exploited to simplify and optimize state checkpointing. To this end, TreeSLS is proposed as a whole-system persistent microkernel that simplifies the whole-system state maintenance to a capability tree and a failure-resilient checkpoint manager. TreeSLS further exploits the emerging non-volatile memory to minimize checkpointing pause time by eliminating the distinction between ephemeral and persistent devices. With efficient state maintenance, TreeSLS also proposes delayed external visibility to provide transparent external synchrony with little overhead.

Evaluation on microbenchmarks and real-world applications (e.g., Memcached, Redis and RocksDB) show that TreeSLS can complete a whole-system persistence in around 100 𝜇s and even taking a checkpoint every 1 ms with reasonable overhead to applications.

This paper represents IPADS‘s endeavor to achieve transparent whole-system-persistence in a custom microkernel operating system. Furthermore, this work follows in the footsteps of ZoFS [SOSP'19] (a high-performance user-space NVM file system), MT^2 [FAST'22] (bandwidth monitoring on NVM/DRAM hybrid platforms), and HTMFS [FAST'22] (a strongly consistent user-space NVM file system on the eADR platform), in exploring further efficient utilization of novel storage devices.

About SOSP

The ACM Operating System Principles Conference (SOSP) and the USENIX Operating System Design and Implementation Symposium (OSDI) stand as premier academic gatherings in the field of operating systems. A total of 229 papers were submitted at this SOSP conference, 43 of which were accepted, and 3 of them won the best paper award.

Paper information

Fangnuo Wu, Mingkai Dong, Gequan Mo, Haibo Chen. TreeSLS: A Tree-structured Microkernel with Efficient Whole-system Persistence on NVM. Proceedings of the 29th ACM Symposium on Operating Systems Principles (SOSP'23). October 2023.

About the Author

The first author of the paper, Fangnuo Wu, is a second-year Ph.D. student in the School of Software, School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University. Her main research interests include operating system design, file and storage systems. Her work has been published in top-tier international conferences in the field of operating systems such as FAST and SOSP. She has been honored with distinctions such as Outstanding Graduate of Shanghai Jiao Tong University and the National Scholarship for Graduate Students.

The corresponding author of the paper, Mingkai Dong, serves as an Assistant Researcher at Institute of Parallel and Distributed Systems (IPADS). His primary research areas include operating systems, file and storage systems, and DNA storage. His work has been published in top-tier international conferences in the field of operating systems such as FAST, USENIX ATC, and SOSP. Notably, the high-performance persistent memory file system SoupFS has been integrated into the OpenEuler Innovation Edition. EROFS, a file system he collaborated on with Huawei, has been merged into the Linux mainline and has become the official file system for Android system partitions.

The other corresponding author of the paper, Haibo Chen, is a distinguished professor at Shanghai Jiao Tong University, where he founds and directs the Institute for Parallel and Distributed Systems (IPADS). His main research areas are operating systems, distributed systems and the application of formal methods. He received Best Paper Awards from SOSP, ASPLOS, EuroSys and VEE, Test of Time Award from DSN, Best Paper Honorable Mention and Research Highlight Award from SIGMOD, Honorable Mention of The Dennis M. Ritchie Thesis Award (Advisor) from SIGOPS. He currently chairs ACM SIGOPS, serves on the editorial board member of contributed articles and co-chairs the Regional Special Sections of Communications of the ACM, co-chair the program committee of EuroSys 2025, and chairs the inaugural technical steering committee of OpenHarmony, an open-source operating system deployed on hundreds of millions of devices. He is an ACM Fellow and IEEE Fellow.