SOLAR Lab’s OSDI paper titled “MemLiner: Lining up Tracing and Application for a Far-Memory-Friendly Runtime” was selected for a Jay Lepreau Best Paper Award at OSDI 2022
Far-memory techniques that enable cloud applications to use remote memory are increasingly appealing in modern data centers, supporting applications’ large memory footprint and improving machines’ resource utilization. Unfortunately, most far-memory techniques focus on OS-level optimizations and are agnostic to managed runtimes and garbage collections (GC) underneath applications written in high-level languages. As a result, when applications have different object-access patterns, GC can severely interfere with existing far-memory techniques and break remote memory prefetching algorithms, causing severe local-memory misses.
This paper presents MemLiner, a runtime technique that improves the performance of far-memory systems by “lining up” memory accesses from the application and the GC so that they follow similar memory access paths, thereby (1) reducing the local-memory working set and (2) improving remote-memory prefetching through simplified memory access patterns. The research team implemented MemLiner in two widely-used GCs in OpenJDK: G1 and Shenandoah. Their evaluation, with a range of widely-deployed cloud systems, shows MemLiner improves applications’ end-to-end performance by up to 2.5 times.
This work was co-led by PostDoc Chenxi Wang (who just began his new role as an Associate Professor at the Institute of Computing Technology, Chinese Academy of Sciences) and Ph.D. student Haoran Ma, and co-authored by Shi Liu (Ph.D. student), Yifan Qiao (Ph.D. student), Jonathan Eyolfson (a former Postdoc who just started his assistant professorship at University Toronto), Prof. Shan Lu (our collaborator at University of Chicago), and Prof. Harry Xu who directs UCLA’s Software System Laboratory for Data Analytics and Machine Learning (SOLAR).