International Business Machines Corporation

Benjamin Hartnett

Scot W Dixon

Mark N Weatherill

Ian L Gregor

A method, program product, and system for a transaction parallelization improvement metric calculation includes receiving tracking data. The tracking data includes information about inbound and outbound subtransactions between domains over a time frame. Each domain includes at least one computer. Using the tracking data received, a number of interactions are determined based on a number of the inbound and outbound subtransactions. A total number of domains is determined using received tracking data. A transaction parallelization metric is calculated using the tracking data and a transaction parallelization improvement metric is calculated based on the number of interactions, the number domains, and the transaction parallelization metric.

Determining a transaction parallelization improvement metric

Embodiments of the present invention disclose a method, computer program product, and system for calculating an average parallelization metric. A tracking application receives tracking data that tracks how at least one server processes a plurality of transactions. Each of the plurality of transactions is comprised of a plurality of subtransactions. A parallelization metric is calculated for each of the plurality of transactions and an average parallelization metric is calculated from the calculated parallelization metrics. The parallelization metric for each transactions calculating using the calculated sum of the response times for all of the subtransactions for one transaction, the longest response time for one subtransaction out of response times for all of plurality of subtransaction that comprise the one transaction, and the calculated actual execution time for the transaction based on the time when the transaction starts to when the transaction has been completed.

Growing community of inventors

West Perth, Australia

Benjamin Hartnett