As every aspect of computer design becomes increasingly dependent on the effective performance and interaction of distributed processors, it is clear that as much effort needs to be focused on software resilience as on hardware fault tolerance. The need for software resilience spans the spectrum: servers; server clusters, farms and arrays; connected desktops and mobiles; network switches and routers; multiple processors in embedded systems; and even multicores in embedded system-on-chip (SoC) devices.
“In both replica and state checkpointing, an efficient IPC is critically important,” said Sven Brehmer, the president and CEO at PolyCore Software, “because in such systems, communications that occur between the application and each of the copies the system maintains use an IPC’s message-passing capabilities.”