An MSCF network consists of two or more processors that cooperate in solving tasks. They may have a client/server relationship or they may work together as equal partners. Distributed system functions, in the sense of an "equal partners" cooperation model, are reserved for the XCS network since it has closer coordination. The functions offered by other types of networks are provided for a job submitter by a server in accordance with the first model.
An elementary requirement for several partners to be able to cooperate is an interface via which they can coordinate themselves. As in a one-processor situation, interfaces are needed to serialize accesses to jointly processed objects and for the synchronization of events that have network-wide relevance. This functionality is made available in the XCS network by the distributed lock manager (DLM).
A network-wide uniform time gives the processors in the network a uniform view when time dependent events take place and the sequence of such events when several occur. The "GTIME" interface was extended for this purpose for the XCS network and is implemented by the subsystem XCS-TIME.
Data that is to be processed network-wide is stored in files on shared pubsets and access to this data is coordinated network-wide. This also applies at file level to a shared pubset network (changes to a file can only be made by one processor in the network at any one time). In addition in the XCS network, the shared file system also offers the option of a network-wide simultaneous update of files with block structure at block level and of PLAM libraries at member level. The synchronization at this level is handled by the distributed lock manager.
The network-wide file processing is supplemented by the network-wide file backup and archiving function with HSMS/ARCHIVE. With the network functionality of Concurrent Copy (CCOPY), the same high data availability can be achieved as in a one-processor situation (with parallel backup and processing across a number of processors).
The aim of the BS2000 network functions is to improve, extend and optimize the system’s "performance" and "availability" characteristics. The basic functions are used via various application-oriented products to attain this goal.
Distributed job networks can be defined with AVAS and the processing of these job networks monitored based on the operating system functions for the management and monitoring of jobs in the network. This allows the load to be optimally distributed between the processors of a shared pubset network.