Details active project to illustrate use of MOSIX to create clusters of surplus machines that can be used to provide network services, application resources.
Distributed OS gives one system image to/over a networked cluster of workstations, research project, Berkeley, John Outerhoust group; high file system performance via client and server caching, process migration to use idle machines; testbed for research in file systems (log-structured, striped, RAID), crash recovery.
hierarchically clustered OS implemented on Hector multiprocessor; manages system resources in clusters, using tight coupling within clusters, loose coupling across clusters. Succeeded by Tornado.
Object-oriented operating system for large-scale shared-memory multiprocessors. Minimizes data sharing via careful design: often eliminates data sharing even when sharing appears natural. Successor to Hurricane.
Global Layer Unix; technology trends in high-speed, low-latency LANs suggest convergence of hardware in Massively Parallel Processors, MPPs, and Networks of Workstations, NOWs. Relative to many MPPs, NOWs offer more cost-effectiveness and scalability, and less hardware and software development time. Soon, the main difference between NOWs and MPPs will be OS software.
Mobile agent system for ubiquitous computing; new way to cope with the wide-area, large-scale ubiquitous computing environment of near future. Allows access to personal and collaborative computing services anywhere, via varied desktop and mobile computing devices.
Distributed Unix operating system for Stanford FLASH multiprocessor; survives hardware and software faults, performance near commercial Unix. Arranged as internal system of independent kernels (cells) for reliability, scalability.
Agent-based communication models show great promise to operate in unpredictable, metamorphic environments, such as mobile computing networks. Excellent basic explanations, clear arguments, well reasoned; from Crossroads, ACM student magazine.
Gives distributed OS services needed to build applications that are wide-area, highly available, incrementally scalable, dynamically reconfiguring, with mechanisms for: resource discovery, global namespace, remote process execution, resource management, authentication, security.
Distributed operating system, to be fully object-oriented, preemptively multithreaded/tasked, event/message-driven. Little code yet, but design documents and links are interesting, useful reading.
Computers /
Computer_Science /
Distributed_Computing
|