The size of the tables that your application program accesses has
a significant impact on the performance of the application program.
Consider the following:
- Large row length:
- For sequentially accessed tables that have a large row length because
of many columns (100 or more), you may improve performance by dividing the
tables into several smaller ones, or by creating a view. This assumes that
your application is not accessing all of the columns. The main reason for
the better performance is that I/O may be reduced because you will get more
rows per page. Splitting the table will affect applications that access all
of the columns because they will incur the overhead of joining the table back
together again. You must decide where to split the table based on the nature
of the application and frequency of access to various columns.
- Large number of rows:
- If a table has a large number of rows, construct your SQL statements
so that the Optimizer uses an index
to access the table. The use of indexes is
very important for achieving the best possible performance.