Xml and Python: Within the Python programming language, several approaches to distributed code management exist. The goal of all such approaches is to allow program logic resident on one server to be utilized by processes on other servers or clients.

---

by David Mertz, PhD, writer and software developer. Intel Corp.

Within the Python programming language, several approaches to distributed code management exist. The goal of all such approaches is to allow program logic resident on one server to be utilized by processes on other servers or clients. The zero-case of such remote invocation is Python's inherent facility for dynamic reloading of support modules, and this provides a basis for understanding further techniques. Remote, dynamic reloading is discussed in this installment, and accompanied by source code examples.

In the paradigm of stand-alone personal computing, a user's workstation contains a number of resources that are used to run an application: disk storage for programs and data; a CPU; volatile memory; a video display monitor; a keyboard and pointing device; perhaps peripheral I/O devices like printers, scanners, sound systems, modems, game inputs, and so on. Even since the 1980s, it has been common for such personal computers to also have network capabilities, but a network card has largely been just another sort of I/O device in the "traditional" paradigm.

"Distributed computing" is a buzz phrase that has something to do with providing more diverse relationships between computing resources and actual computers. Different resources can enter into different sorts of relationships—some hierarchical, others arranged in lattices, rings, and various other topologies. Some of many possible examples are SANs (storage area networks) centralize persistent disk resources for a large number of computers; in the opposite direction, file-sharing protocols like Gnutella and Freenet decentralize data storage and its retrieval; the X-Window System and VNC (AT&T's Virtual Network Computing) allow display and input devices to connect to physically remote machines; protocols like Linux Beowulf allow many CPUs to share the processing of a complex computation, while projects like SETI@Home (NASA's Search for Extraterrestrial Intelligence), GIMPS (Great Internet Mersenne Prime Search), and various cryptographic "challenges" do the same with much less need for coordination.

A few years ago mostly before the explosive growth of the Internet—terms used to describe redistribution of resources were client/server and n-tier architecture (both focusing more on hierarchical relations). The Internet, the Web, and the public consciousness of both have changed the words, and have also shifted emphasis toward graphs, away from trees.