Advanced Data Management

10 Distributed Database Systems

For several decades, centralized database management systems – running on a single database server – have been predominant. There are several reasons for this:

– complexity of a single-server system was lower and administration easier;

– the prevalent use case was to evaluate short queries (frequent reads) on a coherent set of data, whereas data modifications happened only rarely (infrequent writes);

– network communication speed was slow and hence sending data between different servers was too costly;

– parallelization required rewriting a query into subqueries and recombining the results and this overhead diminished the positive effects of a parallel execution of subqueries.

Whenever there were more demanding ...