OLTP Applications

OLTP applications are characterized by quick, indexed access to a small number of data items. An order-entry system is an example of a typical OLTP system. The transactions handled by OLTP applications are usually simple and predefined.

OLTP applications can be characterized as follows:

• Simple transactions that involve small amounts of data
• Indexed access to data
• Many users
• Frequent requests
• Very fast response times

The default behavior of the database server is ideal for OLTP transactions, optimizing performance for short transactions that require rapid response times. All queries have the same priority for CPU, memory, and disk I/O.

Queries that require quick response and generate only a small amount of information should not use PDQ. For example, the following queries should not use PDQ:

• Do we have a hotel room available in Berlin on December 8?
• Does the store in Mill Valley have green tennis shoes in size 4?

The impact of PDQ on OLTP queries can be dramatic. One PDQ parameter limits the number of simultaneous queries the database server can perform. Suppose the number of simultaneous queries is set to 4. If another query requests service, it must wait until one of the previous four queries finishes. If the four queries are decision-support queries, the delay could be several minutes. Typical OLTP queries must be processed immediately.