Can fault-tolerant systems be set up on embedded systems?
The Polyhedra database provides a journalling mechanism that allows database changes to be both persisted to non-volatile store (such as disk or flash-memory file systems) and replicated to a standby database to provide a fault tolerant database service. Where there is no built-in file system, Polyhedra provides various options for storing the persistent data: • The RTRDB can use Flash Memory to keep up-to-date snapshots of the system; • The RTRDB can do a warm restart (typically from battery backed RAM), where it looks at memory (reserve allocation pool) to see if it can find a valid copy of its data structures from a previous run of the system; • The RTRDB can use a remote filing system or ftp to keep snapshots of the database; There can be a standby database set up on a remote machine. This is fed with the changes to all data (not just the persistent data) to keep it up to date. Clients may make connections that will connect to the current master database, and automatically fail ove
