Home > MOSIX

It was developed since 1981 at the Hebrew University of Jerusalem by Prof. Amnon Barak and others. It was first called MOS (Multicomputer OS), and was running on BSD.

In 1999 it was ported to LINUX.

The core of MOSIX are adaptive algorithms that monitor and respond to cluster-wide resource requirements (of all the processes) by automatic migration of processes to the best available nodes. These algorithms are geared for ease-of-use, maximal performance and overhead-free scalability.

The algorithms of MOSIX use preemptive (transparent) process migration for:

• Automatic distribution and redistribution of processes;
• Moving processes from slower to faster nodes;
• Load balancing - for even work distribution;
• Migrating processes from nodes that run out of memory, to prevent swapping or thrashing;
• Parallel I/O - by migrating intensive I/O processes to file servers (rather than the traditional way of bringing data to the processes).

Recent trends in cluster and grid computing, in which many applications consist of large number of processes and/or large amounts of data, e.g., genomic or financial modeling, prompt the development of the Mosix Parallel I/O (MOPI) package. MOPI enhances the computing capabilities of Mosix with the ability to support massive parallel I/O. It is targeted for applications that need to process vast amounts of data, ranging from few GByte/s to Terabyte/s.

openMosix

After a disagreement about the future of MOSIX, Moshe Bar decided to fork and started the openMosix project on February 10, 2002.