gtpg1m09General Information

TPF System Characteristics

The TPF system can help your organization give customers prompt and accurate service by providing:

Message-Driven Processing

The TPF system is a message-driven system; each query that the TPF system receives is called a message. A transaction refers to the series of message pairs (input/reply) necessary for completing a customer's request. A business transaction can consist of multiple message pairs such as a cash withdrawal from an automated teller machine (ATM), or an airline reservation with advance seating. The number of message pairs required to complete a transaction depends on the application design and the complexity of the specific information required by the agent. For example, a credit verification often results in a transaction of just one message pair.

Very High Throughput with Quick Response Time

The TPF system provides very high throughput measured by the number of messages per second. A single transaction usually involves several messages and multiple database accesses.

The typical TPF system processes several hundred messages per second. For even higher throughput, a loosely coupled TPF system can process thousands of messages per second.

You can obtain more than 5000 messages per second depending on your configuration. The maximum throughput in a fully configured loosely coupled complex is unknown because no tool exists that is capable of testing the TPF system's maximum performance. Your system's actual capacity with the TPF system depends on the hardware configuration and the application design.

The response time of the TPF system within a network is typically less than three seconds from the time the user sends a message to the time the user receives a response to the message. During that 3 seconds, the message typically spends approximately 0.5 seconds in the TPF system.

High Availability

The TPF system provides very high system availability. In many cases, the scheduled availability has exceeded 99.9%, which represents fewer than 10 minutes of downtime per week. Scheduled availability is measured in terms of a system processing work 24 hours per day, 7 days per week, minus the time for scheduled outages. The TPF system can restart in one-half minute to 2 minutes of a system failure, so the end user is often not even aware of the failure.

Even higher availability can be achieved when TPF central processing complexes (CPCs) are run in a loosely coupled configuration. The system can be designed so that the failure of any one CPC will be undetected by most end users. Additional CPCs can be brought into the online complex without any outages.

Having a properly configured backup system or backup processor can increase the scheduled availability to greater than 99.9%. TPF customers who have single central processing complex (CPC) systems usually have another CPC that is capable of taking over from the primary one. Usually, this backup system is assigned to run jobs that can be interrupted if necessary. Typically, an installation can switch over to the backup CPC within minutes of the primary CPC failure.

Data Integrity

The introduction of virtual address spaces in the TPF environment has significantly increased the integrity of the data environment in the TPF system.

The TPF system also provides the basic tools your installation might need for additional data integrity and recovery. The TPF system reaches its high transaction rate in part by allowing installations to determine the degree of integrity needed for each system resource and to build their own recovery.

Low Cost Per Transaction

Very short path lengths for critical system services such as DASD I/O help to make efficient use of the TPF system's resources and to provide a very low cost per transaction.