IBM Books

Administration Getting Started


Separating Different Types of Data

One aspect of physical database design you might consider is to keep indexes and large object (LOB) data separate from the rest of the table data when you create a table. This option is described in "Step 1. Creating a Table". These separate table spaces can possibly be on different media (containers) than the rest of the data. This flexibility can be used to increase database performance and availability. You can also allocate a table space to a specific buffer pool. See "Buffer Pool" for a description of a buffer pool.

Indexes stored in a different table space from that used to store other table data can allow for more efficient use of disks by reducing the movement of read/write heads. You can also associate index table spaces with faster devices, which can speed up index access.

If you store long data types and LOB data in a different table space, you can allocate a special device to this type of data. For example, you may choose to store LOB data on devices that have a very large capacity but are slow, and put the rest of your data, which may be accessed more often, on a faster device. You can turn off logging of individual LOB columns for improved performance.


[ Top of Page | Previous Page | Next Page | Table of Contents | Index ]

[ DB2 List of Books | Search the DB2 Books ]