Calculating Network Response Times

Response time monitoring calculates host and network response times. Host response time is the time that is taken by the application to process a transaction. To calculate response times, ULTRAOPT invokes definite response for the application. (Definite response is a VTAM function which specifies that traffic sent to the terminal be acknowledged.)

Network response time represents the time that is required for the output of a transaction to travel through the network to the terminal, plus the time that is required for the definite response to return to the VTAM application. Response times can only be monitored for LU2 sessions and optimized LU0 and LU1 sessions. A definite response cannot be forced on sessions that use LU 6.2 protocol, nor can it be used for bisynchronous and 4700/3600 (SNA financial) terminals.

Consider the following variables:

• Time A: Transaction flows from the terminal through the network to the host.
• Time B: Transaction is processed in the host.
• Time C: Transaction output flows from the host through the network to the terminal.
• Time D: Definite response is sent from the terminal (PU) to acknowledge receipt of output.

An application user typically perceives response time to be host response time added to the network response time. Users typically define response time as Time A + Time B + Time C.

Response time, as calculated by the subsystem RTM function, is Time B + Time C + Time D.

The subsystem assumes that Time A is equivalent to Time D. The size of transaction input and line speed are determining factors in response time calculations.