1.34k likes | 1.69k Views
Distributed Databases. Semantic Data Control. Distributed Query Processing. Distributed Database Design. System Architecture. Distributed Transaction Management. ANSI/SPARC Architecture DBMS Implementation Alternatives Datalogical Architecture Clients/Server Architecture
E N D
Distributed Databases Semantic Data Control Distributed Query Processing Distributed Database Design System Architecture Distributed Transaction Management Yangjun Chen
ANSI/SPARC Architecture DBMS Implementation Alternatives Datalogical Architecture Clients/Server Architecture Components of a Multi-DBMS System Architecture Yangjun Chen
Relation Fragmentation • Horizontal, Vertical, Hybrid • Correctness of Fragmentation • Completeness, Reconstruction, Disjointness • Algorithms • COM_MIN Al. , Bond Energy Al. • Fragment Allocation Distributed DBMS Design Yangjun Chen
Algorithms for horizontal fragmentation: • COM_MIN Algorithm Given a relation and a set of predicates, it will find a complete and minimal set of predicates • Horizontal Algorithm Given a complete and minimal set of predicates, it will produce a horizontal fragmentation Yangjun Chen
Global affinity measure: GAM = aff(Ai, Aj)*[aff(Ai, Aj-1) + aff(Ai, Aj+1) + aff(Ai-1, Aj) + aff(Ai+1, Aj)] Since the affinity matrix is symmetric, we have GAM = 2 * aff(Ai, Aj)*[aff(Ai, Aj-1) + aff(Ai, Aj+1 )] AM = aff(Ai, Aj)*[aff(Ai, Aj-1) + aff(Ai, Aj+1 )] Yangjun Chen
Global affinity measure: AM = aff(Ai, Aj)*[aff(Ai, Aj-1) + aff(Ai, Aj+1)] = [aff(Ai, Aj)*aff(Ai, Aj-1) + aff(Ai, Aj)*aff(Ai, Aj+1)] = [ aff(Ai, Aj)*aff(Ai, Aj-1) + aff(Ai, Aj)*aff(Ai, Aj+1)] = [bond(Ai, Aj-1) + bond(Ai, Aj+1)], Where bond(Ax, Ay) = aff(Az, Ax)*aff(Az, Ay). Yangjun Chen
Relation schema: R[A1, …, Ai, …, Aj, …, Ak-1, Ak, …, An] Affinity matrix: A1, …, Ai, …, Aj, …, Ak-1, Ak, …, An … … … … … Clustered affinity matrix: Ai1, …, Ail,Ai , Aj, Ai(l+3), …, Ai(k-1) Ai1, …, Ail,Ai , Ak , Aj, Ai(l+3), …, … … … … … … new old Yangjun Chen
AMold = [bond(Ais, Ai(s-1)) + bond(Ais, Aj(s+1))] + bond(Ai, Ail) + bond(Ai, Aj) + bond(Aj, Ai) + bond(Aj, Ai(l+3)) + [bond(Ais, Ai(s-1)) + bond(Ais, Aj(s+1))] AMnew = [bond(Ais, Ai(s-1)) + bond(Ais, Aj(s+1))] + bond(Ai, Ail) + bond(Ai, Ak) + bond(Ak, Ai) + bond(Ak, Aj) + bond(Aj, Ak) + bond(Aj, Ai(l+3)) + [bond(Ais, Ai(s-1)) + bond(Ais, Aj(s+1))] AMnew – AMold = 2* bond(Ai, Ak) + 2* bond(Aj, Ak) – 2* bond(Ai, Aj) Yangjun Chen