(a) providing a user interface for running an experiment to analyze performance of an application program execution, the application program being executed by the multiple nodes of the clustered system, wherein each of the multiple nodes include a trace collector for collecting latency data, bandwidth data and communication data and each trace collector data is sent to a central trace collector for analysis;

(b) selecting the experiment from among a plurality of options; the plurality of options providing a user an option to select from (i) an experiment that the user has previously run; (ii) start a new experiment; or (iii) select a template to run a series of experiments under different conditions;

(c) based on user selection, executing the application program at the multiple nodes in the clustered system; wherein portions of the application program are executed simultaneously by the multiple nodes;

(d) gathering trace data for the multiple nodes in the system, the trace data including data in accordance with communication between at least two nodes in the system;

(e) analyzing the gathered trace data to determine impact of latency, bandwidth and communication on a total effect of the selected experiment and to determine a duration that the application program spends in performing computing functions and a duration that the application program spends on an interface call used for passing messages between the multiple nodes executing portions of the application; and

(f) displaying data for latency, bandwidth, and communication; wherein the displayed data for communication includes a communication number that shows a percentage effect that communication had on the total effect of the selected experiment.