Web site: (not active)
Origin: USA
Category: Microkernel, Others
Desktop environment: CLI
Architecture: unknown
Based on: Independent
Wikipedia: Hydra_(operating_system)
Media: Install
The last version | Released: 1975
Hydra (C.mmp/Hydra) – a capability-based, object-oriented microkernel operating system, created at Carnegie-Mellon University, as part of the C.mmp.
In the early 197Os,a project began at Carnegie-Mellon University to investigate computer structures for artificial intelligence applications. These applications required substantial processing power available only on costly high-performance processors. At that time, however, relatively inexpensive mini-computers were becoming available. Therefore, the project sought to demonstrate the cost performance advantages of multiprocessors based on inexpensive minicomputers.
Included are detailed descriptions of the PMS structure of C.mmp (a multiprocessor built from minicomputers) and its major components. An overview of its operating system, Hydra, is provided with emphasis on those sections most concerned with and influenced by the architecture. The project is also discussed in terms of performance, reliability, programming methodologies, and problems encountered.
In 1971 the Computer Science Department at Carnegie-Mellon University (CMU) undertook a project to construct C.mmp (Computer. multi-mini-processor), a relatively large-scale multiprocessor, from minicomputers. A number of project goals and criteria influenced the design:
– Minicomputers would be used as the processing elements of a multiprocessor that would support a general-purpose, time-shared environment.
– The machine would be symmetric: there would be no master-slave relation among the processors.
– A large address space would be provided.
– As much commercially available hardware as possible would be used.
Two fundamental design decisions that permit experimentation on the Hydra system are:
– the separation of policy and mechanism in the kernel [Levin 751,and
– the use of an object-based model of computation with capability protection
Main features:
– Separation of policy and mechanism: a kernel of mechanisms of “universal applicability” would be created from which varying policies could be implemented.
– A capability-based protection system and an object-oriented virtual memory would provide support for data abstraction; it would be extensible to user-defined data-types.
– The software would exploit the existence of multiple copies of many hardware elements for reliability.
– The structure of the system would be non hierarchical.
– The system would be able to run for extended periods with no human operator.
The project was under active development between 1971 and 1975.
