This invention generally relates to computer systems and architectures, methods and computer program code for increasing the robustness of process control and manufacturing automation systems and the like, in particular to provide improved handling of error and/or failure conditions. We describe an architecture for an OPC Failover system, the architecture comprising: a plurality of OPC servers; at least one OPC client; a computer network linking said OPC servers and said OPC client; and an OPC Failover service coupled to said network and not coupled between said OPC client and said OPC servers, said OPC Failover service being configured to monitor at least one of said OPC servers for an error condition and to redirect network communications of said OPC client from one of said servers to another on detection of said error condition.