1、1 7. Distributed Database Systems Principles of Database Systems, Xu Lizhen 2 7.1 Introduction What is DDB? A DDB is a collection of correlated data which are spread across a network and managed by a software called DDBMS. Two kinds: (1) Distributed physically, centralized logically (general DDB) (2
2、) Distributed physically, distributed logically too (FDBS) We take the first as main topic in this course. Principles of Database Systems, Xu Lizhen 3 Distribution Correlation DDBMS Features of DDBS : Principles of Database Systems, Xu Lizhen 4 The advantages of DDBS: Local autonomy Good availabilit
3、y (because support multi copies) Good flexibility Low system cost High efficiency (most access processed locally, less communication comparing to centralized database system) Parallel process The disadvantages of DDBS: Hard to integrate existing databases Too complex (system itself and its using, ma
4、intenance, etc. such as DDB design) Principles of Database Systems, Xu Lizhen 5 The main problems in DDBS: Compared to centralized DBMS, the differences of DDBS are as follows: Query Optimization (different optimizing goal) Concurrency control (should consider whole network) Recovery mechanism (all
5、sub-transactions must commit or abort simultaneously) Another problem specially for DDBS: Data distribution Principles of Database Systems, Xu Lizhen 6 7.2 Data Distribution 7.2.1 Strategies of Data Distribution (1) Centralized: distributed system, but the data are still stored centralized. It is si
6、mplest, but there is not any advantage of DDB. (2) Partitioned: data are distributed without repetition. (no copies) (3) Replicated: a complete copy of DB at each site. Good for retrieval-intensive system. (4) Hybrid (mix of the above): an arbitrary fraction of DB at various sites. The most flexible
7、 and complex distributing method.2 Principles of Database Systems, Xu Lizhen 7 Comparison of four strategies 1234 flexibility complexity Advantage of DDBS Problems with DDBS Principles of Database Systems, Xu Lizhen 8 7.2.2 Unit of Data Distribution (1) According to relation(or file), that means non
8、 partition (2) According to fragments Horizontal fragmentation: tuple partition Vertical fragmentation: attribute partition Mixed fragmentation: both Principles of Database Systems, Xu Lizhen 9 The criteria of fragmentation: (1) Completeness: every tuple or attribute must has its reflection in some
9、fragments. (2) Reconstruction: should be able to reconstruct the original global relation. (3) Disjointness: for horizontal fragmentation. Principles of Database Systems, Xu Lizhen 10 7.2.3 Problems Caused by Data Distribution 1) Multi copies consistency 2) Distribution consistency Mainly the change
10、 of tuples store location resulted by updating operation. Solution: (1) Redistribution After Update: Select-Move-Insert-Delete (2) Piggybacking Check tuple immediately while updating, if there is any inconsistency it is sent back along with ACK information and then sent to the right place. Principle
11、s of Database Systems, Xu Lizhen 11 3) Translation of Global Queries to Fragment Queries and Selection of Physical Copies. 4) Design of Database Fragments and Allocation of Fragments. Above 1)3) should be solved in DDBMS. While 4) is a problem of distributed database design. Principles of Database S
12、ystems, Xu Lizhen 12 7.3 Federated Database In practical applications, there are strong requirements for solving the integration of multi existing, distributed and heterogeneous databases. The database system in which every member is autonomic and collaborate each other based on negotiation - federa
13、ted database system. No global schema in federated database system, every federated member keeps its own data schema. The members negotiate each other to decide respective input/output schema, then, the data sharing relations between each other are established.3 Principles of Database Systems, Xu Li
14、zhen 13 The schema structure in federated database System FS 1 FS 3 FS 2 IS 3 IS 2 IS 1 ES 3 ES 2 ES 1 CS 3 CS 2 CS 1 DB 1 DB 2 DB 3 Site1 Site3 Site2 Principles of Database Systems, Xu Lizhen 14 FS i = CS i + IS i FS i is all of the data available for the users on site i . IS i is gained through th
15、e negotiation with ES j of other sites (j i). Users query on FS i the sub-queries on CS i and IS i the sub-queries on corresponding ES j . The results gained from ES j the result forms of corresponding IS i , and combined with the results get from the sub-queries on CS i , then synthesized to the ev
16、entual result form of FS i . Principles of Database Systems, Xu Lizhen 15 7.4 Query Optimization in DDBMS Optimization goal: minimize the transmission cost on network Algebra optimization Translation of global queries to fragment queries and selection of physical copies Query Decomposition Global qu
17、ery plan Principles of Database Systems, Xu Lizhen 16 An example of global query optimization Global query optimization may get an execution plan based on cost estimation, such as: (1) send R2 to site1, R (2) execute on site1: Select * From R1, R Where R1.a = R.b; R1 R2 Site1 Site2 Select * From R1,
18、R2 Where R1.a = R2.b; Principles of Database Systems, Xu Lizhen 17 7.5 Recovery Mechanism in DDBMS The basic principle is the same as that in centralized DBMS Distributed transactions : the key of distributed transaction management is how to assure all sub-transactions either commit together or abor
19、t together. Realize the sub-transactions harmony with each other relies on communication, while the communication is not reliable. Two phase commitment protocol Combination of failures Principles of Database Systems, Xu Lizhen 18 7.6 Concurrency Control in DDBMS The basic principle is the same as that in centralized DBMS, demand concurrent transactions to be scheduled serializably Because of multi copies, need locking globally Communication overhead Global deadlock
