Dealing with failures in heterogeneous databases

The distributed information processing scenario has changed dramatically over the last few years. While there is increased pressure on individuals to obtain information about a wide range of topics, the databases where that information lies exist on a heterogeneous collection of database management systems, residing on a variety of networks, each under the control of a different organization, and with very little standardization among the DBMSs. This situation has motivated much recent research on what are variously called heterogeneous databases, federated databases or multidatabases. The authors address the issue of failures in such heterogeneous environments. They start out by briefly recounting some of the failures that they know how to resolve in both centralized and distributed databases. Then they point out some of the salient features of federated environments and show how those features make possible new kinds of failures. Finally, they present some possible solutions to these new types of failures