A fault-tolerant real-time commercial LAN

Much of the communication in hard real-time systems requires predictable, low latency message delivery high average throughput; unfortunately LAN technologies are usually measured by the latter. A further departure from traditional LAN usage is that mission critical messages are many times very small but coexist with extremely large data on the same network. Optimizing protocols designed to move large volumes of data to provide real-time performance for small, critical messages has consumed millions of dollars. This paper focuses on the reverse-using a protocol designed for small messages and extending it for large pieces of data.