Performance of the MOSIX Parallel System for a Culster of PC's

摘要

The scalable PC scluster at Hebrew University consists of 48 Pentiumand Pentium-Pro servers that are connected by fast Ethernect and the Myrinet LANs. It is running the MOSIX operating system, an enhancement of BSK/OS with algorithms for dynamic resource sharing that are geared for performance scalability in scallable computing chuster. These algorithms use a preemptive process migration mechanism for load-balancing and memory-sharing, in order to create a convenient multi-suer time-sharing execution environment for HPC, particularly for applications that are written in PVM or MPI. This paper gives a brief overview of MOSIX ans its resolurce sharing algorithms. Then the paper presents the performance of these algorithms as well as the performance of several large-scale, parallel applications.