Reliable transaction execution in multidatabase systems

The issue of executing concurrent transactions reliably in a multidatabase system is addressed. The authors´ formalization assumes the most pessimistic scenario where individual database managers that make up the multidatabase system are totally autonomous. The authors present algorithms that ensure transaction atomicity and crash recovery. The atomicity algorithms are based on the emulation of the well-known two-phase commit protocol. It is applied among autonomous participants that neither know of each other´s existence nor how to communicate