SOSP '95 Proceedings of the fifteenth ACM symposium on Operating systems principles
Exploiting deferred destruction: an analysis of read-copy-update techniques in operating system kernels
Improving performance by embedding HPC applications in lightweight Xen domains
Proceedings of the 2nd workshop on System-level virtualization for high performance computing
Factored operating systems (fos): the case for a scalable operating system for multicores
ACM SIGOPS Operating Systems Review
The multikernel: a new OS architecture for scalable multicore systems
Proceedings of the ACM SIGOPS 22nd symposium on Operating systems principles
Helios: heterogeneous multiprocessing with satellite kernels
Proceedings of the ACM SIGOPS 22nd symposium on Operating systems principles
NOVA: a microhypervisor-based secure virtualization architecture
Proceedings of the 5th European conference on Computer systems
Tessellation: space-time partitioning in a manycore client OS
HotPar'09 Proceedings of the First USENIX conference on Hot topics in parallelism
An analysis of Linux scalability to many cores
OSDI'10 Proceedings of the 9th USENIX conference on Operating systems design and implementation
| Hi-index | 0.00 |
With a steady trend from singe-core to multicore processors, scalability has become a significant design issue for the Operating Systems (OS), as many critical OS functions must be re-designed in order to achieve scalable performance. While numerous efforts have been made to improve scalability of monolithic OS kernels, comparatively little work has been done for microkernels. In this paper, we begin by studying the scalability of Fiasco.OC, a state-of-the-art microkernel implementation. We then present OmniRE, a new personality for the Fiasco.OC microkernel that is aimed at being multicore scalable. Compared to L4Re (the vanilla "off-the-shelf" Fiasco.OC personality), OmniRE aims to eliminate contention by decentralizing resource management, scheduling, and kernel access. The design also aims to minimize inter-process communication (IPC) across CPUs by localizing resource functionality such as page-fault handling. We conduct experiments to compare OmniRE against L4Re as well as Linux on a 48-core AMD server and a 6-core Intel workstation. Our results indicate that OmniRE provides better scalability than L4Re and can in fact exceed absolute performance of Linux in memory page management at higher core counts.