Every time you run a program, be it interactive or a system service, you create a process. A process is a piece of code being executed by the operating system. A process can spawn child processes which can spawn new threads . In layman's terms, a program can start other instances of itself or another program and they, in turn, can start small pieces of executable code which can run in parallel with the main program.

Programs do not start spontaneously. Someone has either got to start them, or instruct the system to start them when some criteria are met. This sentence is the acknowledgement of the simplicity behind a computer system; it can't think on its own, humans have to tell it what to do one way or the other. Based on how a program starts, it process will be owned by some user.

In the first and simplest case, when you start a program, the ownership is almost self-understood. You are logged in as some user, so the process of the program you have executed is owned by your user. It's simple as that. This also implies that the process has the same permissions as the owning user, that's why we say that the process runs under this user; its access level is at most as much as the owning user, so the process is under the user.

The other case, instructing the system to start a process, is somewhat different. Usually, the utilities which are used to start programs automatically are the system initialisation scripts, time-based execution programs (for example, cron and at ), etc. All of these programs are in most cases owned by root and are executed under root privileges. On top of that, most programs started this way are system services, running as long as the system is up and running. But do you remember what we said before? Root is the God of the system. Normally, these programs would get root's privileges, posing a huge security hole. If there is a bug in the program and some malicious user exploits it, he could wreck havoc on the system; root is above all restrictions.

In order to combat this possibility, UNIX systems employ a feature which allows processes to drop privileges and run under a different user than the one which started them. In fact, they change their ownership! To prevent abuse of this feature, a process must run under root privileges to be able to switch to another user. This feature is extensively used by system services, including MySQL and Apache.

In the context of web site management, Apache is of special interest. Apache is the de facto web server for Linux systems and is being used by over 50% of Internet sites, according to NetCraft's August 2008 survey. Chances are you are using it on your site, too. Apache, like most UNIX services (affectionately called daemons) uses the feature to drop privileges. The user and group under which it runs are defined in its configuration files. These configuration files are usually out of the reach of regular users (like you!) on commercial hosts, for security reasons.

There is a special case which acts as the exception to the Apache rule. Many commercial hosts run suPHP-enabled Apache installations. This is an extension to the normal PHP's mode of operation which allows each PHP page to run in a process owned by the file's owner (more on file ownership in the next sub-section). This means that each of the PHP files under your account on such a host run as the user which has been assigned to your account. And, if this still isn't apparent to you, such hosts nullify the burden of ownership and permissions (more on permissions in the next section). To put it clearly: with suPHP the file owner, your own user and the Apache user are one and the same. If you are looking for a decent host, find one which is using suPHP. It's better for security and removes a lot of administrative burden from you. A win-win situation.