MDC addresses these drawbacks. Data is clustered physically in multiple 

dimensions. The indexes for each dimension are not record based but block 

based, thus reducing their size (and effort needed for logging and maintaining) 

dramatically. Reorganization of the table in order to re cluster is not necessary.

At the bottom line, MDC provides a powerful method for improving the 

performance of 

SELECT

, 

INSERT

, 

UPDATE

, and 

DELETE

 statements.

Example 10 10 shows the 

CREATE TABLE 

statement of an MDC table clustered in 

two columns: 

Region

 and 

Year

. The block indexes for each dimension are created 

automatically.

Example 10 10   Creating an MDC table

CREATE TABLE sales ( 

Customer VARCHAR(80), 

Region CHAR(5), 

Year INT

) 

ORGANIZE BY DIMENSIONS (Region, Year);

Figure 10 4 shows the data clustering according to two dimensions as defined in 

Example 10 10. Actually, the number of dimensions for a table is only limited by 

the system size.

Dimension

Dimension

Block Index

Block Index

East, 1993

East, 1996

North, 1996

Region

North, 1997

Year

North, 1998

South, 1999

Extent (Four pages)

Data

Data

Figure 10 4   MDC table and indexes

332 

MySQL to DB2 UDB Conversion Guide