What’s a process
In practice, a process is basically a binary code being executed by processors. Processes are launched by the operating system (since the operating system controls the execution flow) and have several properties: PID (unique identifier of a process); Name; User ID; Command line that was used to launch the process (arguments passed to the binary); Environment; CPU and memory usage; Other various OS-specific properties.
How to identify a process
When you monitor an application, you typically want to check that the application's processes are running properly. The problem lies in how to identify the processes of this application, how to recognize them amongst all of the running processes.
The only thing that really identifies a process is its PID (Process ID). But since the PID is an integer number randomly set upon the process startup, most often you cannot use it to identify the processes of an application (unless the application gives you its PID in a so-called PID file).
In general, you recognize application processes by their name if this criterion is enough to distinguish them from other processes. If the name of the process is not sufficient, you can identify application processes by parsing the process’s command lines. This is typically useful with scripts and java processes, whose process names are the same: java, CSCRIPT.EXE, etc.
Process name
Under Windows, the name of a process is basically the file name of the binary file which is being executed: Java.exe, IisAdmin.Exe. It always includes the ".EXE" extension. Process names can easily be shown in Windows Task Manager.
Under UNIX, the process name could be either the file name of the binary being executed, including the path or not, or something completely different (e.g. Oracle processes).
 The naming of processes is highly platform dependent. Linux processes are not named in the same way as on HP-UX servers, for example. Under UNIX, process names can be shown by executing the "ps –e –o name" command line.
Process command line
Every process is launched through a command line, which consists of the file path to the binary which has to be executed, and arguments that have to be passed to the binary: <path to the binary file> <argument1> <argument2> etc.
If the directory of the binary file is in the PATH environment variable, the path may not be included in the command line: <binary file name> <argument1> <argument2> etc.
This is the only way to distinguish Java processes and scripts from others, because their process names are all identical (Java.EXE). Unfortunately, in Windows, there is no easy way to see the command lines of the currently running processes. Under UNIX, processes command lines can be shown by executing the "ps –e –o comm" command.
Process user ID
On both Windows and UNIX systems, processes run "as" a user. Depending on this, the process may be allowed to access various system resources (files, network, databases, etc.). In secured environments, most applications processes have to run as a specific user to let them access the application resources. If the processes run as another user, the application is very likely to fail and not run properly. This is why it could be important to check that the processes of the application you want to monitor are running as the appropriate user.
PID file
A classic way for applications to indicate they are running is to write the PID of their process into a given file. In this case we only need to read this file and check whether the PID written in the file corresponds to a running process. Please note that now the PID file is not provided for all the applications and most Windows applications do not provide PIDs.
| 
