Block transfers of information in data processing networks

Abstract

Secondary storage subsystems exchange messages and data with host data processing systems and also forward messages between host systems. Host systems thereby communicate with each other in addition to having access to data in subsystem storage. Access to subsystem storage is initiated by a 'request' sent from a host to the subsystem. Each request is a message containing an array of one or more commands, each command specifying a transfer of data or a control function to be performed by the subsystem. A subsystem may process more than one request at a time. It also may process the commands in a request in an arbitrary sequence suited to the availability of subsystem resources and data links to host systems. After all commands in a request have been processed the subsystem transmits an associated 'completion' message to the host system which originated the request. The completion message indicates the status of completion or abnormal termination of each command in the associated request. An 'adapter' processor associated with each host and subsystem operates on an asynchronous basis to transfer messages and data relative to the associated host or subsystem. One or more processing 'engines' in each adapter communicates with one or more CPU's in the associated host or subsystem through an associated 'adapter store'. A portion of each adapter store is used as a buffer pool for constructing 'subchannel control spaces' to control transers of messages and data. Elements of each subchannel control space are returned to free status as soon as they are not needed for sustaining associated transfers.