Collection of timing and coverage data through a debugging interface

Abstract

Techniques for gathering execution information about an application, such as a distributed application, are described. Key communication points in cross execution context calls, such as remote procedure calls, are determined and control is transferred to instrumentation routines to insert and extract execution information. Outgoing remote procedure calls are intercepted on a client that inserts call origin information into the request sent to a server system. Messages received by a server are intercepted. The server system extracts the call origin information and additionally inserts other information in a response sent to the client system upon completion of a remote procedure call. In turn, the client system intercepts the response and extracts other performance information. On each client and server system, information is gathered by a reader and forwarded to a local collector. This information may be further forwarded to and correlated by a client collector from one or more remote server collectors in accordance with processes of each distributed application. Various statistics for a distributed application may be determined in addition to per process statistics. These include wire time, code coverage as related to the distributed application, remote procedure call tracing, and performance profiling. A variety of techniques are described to obtain program execution information in connection with an executing application including instrumentation techniques and use of a debugger interface to obtain profiling and other execution information. All of the program execution data may be collected and correlated at one or more particular points using other techniques described to represent coordinated application monitoring.