DTrace has the ability to follow the operation of applications from within the application source code, through system libraries, through system calls, and into the kernel. This visibility allows the root cause of issues (including performance issues) to be found and quantified, even if it is internal to a kernel device driver or something else outside the boundaries of the application code. Using DTrace, questions such as the following can be answered.
• What transactions are occurring? With what latency?
• What disk I/O is the application performing? What network I/O?
• Why is the application on-CPU?
As an example, the following one-liner frequency counts application stack traces when the Apache Web server (
httpd) performs the ...