Load testing is the process of applying ordinary stress to a software system to determine the system performance under normal conditions. In a typical load testing lifecycle, three kinds of service tools are involved: test case recording service tools that make testers easier to generate test cases through a web browsing-like behavior; test case execution service tools that exercise test cases with simulations of a large number of concurrent users; system resource monitoring service tools that provide information of system footprints during the test case executions. However, using these three kinds of service tools one by one to complete a load testing may require extra effort on operating and configuring each service. In this paper, we proposed a framework for composing the three types of service tools as an integrated service for load testing. A raw test case recorded by Badboy tool is automatically converted into an expanded test case that can be executed by JMeter. JMeter and Cacti are then automatically invoked by the framework. The execution time period of JMeter is automatically identified as the input to Cacti for resource monitoring of the system under test. The test report together with system footprints is also automatically generated. In the experimental evaluation, the result shows that the framework significantly save time on operating and configuring the load testing service tools than the traditional approach under a t-test.