Fault Tolerant Systems
Support provided for supervision in SCIOPTA can also be used in a distributed system. CONNECTOR processes, board support processes and the network system can be supervised by the kernel. Remotely connected processes will be informed by the kernel if resources in the communication path are no longer available.
In a SCIOPTA distributed system there is no need for having a master node. A node or CPU that disappears will not affect the whole distributed system.
SCIOPTA supervision (observations), transparent communication and the module concept allow you to build fault tolerant distributed systems.
CONNECTOR is a SCIOPTA communication process that connects SCIOPTA systems on different CPUs in a distributed environment.
CONNECTOR is the essential part in a distributed Multi-CPU system running SCIOPTA. Each node is controlled by a CONNECTOR process, which has knowledge of distributed processes taking part in the communication system.
CONNECTOR processes are responsible for redirecting messages, of knowing the location of distributed processes and hiding the distributed system from the user programmer.
CONNECTOR processes have an open interface and do not depend on any network topology. Therefore any network stacks (e.g. TCP/IP, CAN) can be used.
A SCIOPTA distributed system is not limited to one type of CPU. Any CPU, where a SCIOPTA kernel and a CONNECTOR process are available, can be included.
The clean message based interface in SCIOPTA allows distributed systems to be designed as quickly and easily as a single CPU system.
Message passing beyond processor boundaries uses the same system calls for transmitting and receiving, although the processes reside on different CPUs.