UMap: Enabling Application-driven Optimizations for Page Management
Authors:
- 1 November 2019
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 71-78
Abstract
Leadership supercomputers feature a diversity of storage, from node-local persistent memory and NVMe SSDs to network-interconnected flash memory and HDD. Memory mapping files on different tiers of storage provides a uniform in- terface in applications. However, system-wide services like mmap are optimized for generality and lack flexibility for enabling application-specific optimizations. In this work, we present UMap to enable user-space page management that can be easily adapted to access patterns in applications and storage characteristics. UMap uses the userfaultfd mechanism to handle page faults in multi-threaded applications efficiently. By providing a data object abstraction layer, UMap is extensible to support various backing stores. The design of UMap supports dynamic load balancing and I/O decoupling for scalable performance. UMap also uses application hints to improve the selection of caching, prefetching, and eviction policies. We evaluate UMap in five benchmarks and real applications on two systems. Our results show that leveraging application knowledge for page management could substantially improve performance. On average, UMap achieved 1.25 to 2.5 times improvement using the adapted configurations compared to the system service.Keywords
This publication has 9 references indexed in Scilit:
- SAGE: Percipient Storage for Exascale Data Centric ComputingParallel Computing, 2019
- MPI windows on storage for HPC applicationsParallel Computing, 2018
- Efficient Memory-Mapped I/O on Fast Storage DeviceACM Transactions on Storage, 2016
- Let's Talk About Storage & Recovery Methods for Non-Volatile Memory Database Systems2015
- On the Role of NVRAM in Data-intensive Architectures: An Evaluation2012
- Benchmarking cloud serving systems with YCSB2010
- Optimization via Reflection on Work Stealing in TBB2008 IEEE International Symposium on Parallel and Distributed Processing, 2008
- Cashmere-VLM: Remote memory paging for software distributed shared memory2003
- Application-controlled demand paging for out-of-core visualization2002