Partial Replication

Definition

A replicated database consists of a set of nodes $\mathcal {N}$ (database servers) and each logical data item x has a physical copy on a subset ${\mathcal {N}}_x$ of the nodes in $\mathcal {N}$. The replication degree $r_x = |{\mathcal {N}}_x|$ of a data item is the number of copies it has. Using full replication, each logical data item has a copy on each of the nodes, i.e., for each data item x of the database, ${\mathcal {N}}_x = {\mathcal {N}}$. Whenever there is at least one data item that does not have copies at all nodes, one refers to a partial replication architecture.

Main Text

Full replication is expensive in update intensive environments as the updates have to be typically executed on all replicas limiting scalability and requiring costly coordination. Thus, the lower the replication factor, the better the potential for scalability and fast update operations. On the other hand, more replicas offer better scalability for read operations, replicas at...