The correct answer is 'Response time,' as it is a primary indicator of application performance that measures the time taken for a system to process an event from the moment of a request by an end user or system process. High response times can lead to slow performance and a negative user experience. While 'Resource allocation' is related to the provisioning of computing resources and 'Network uptime' indicates network reliability, neither directly measure application performance. 'Cost per transaction' deals with financial metrics and not performance. 'Physical server temperature' is not applicable in a cloud context where hardware is abstracted from the user.