CAVEAT: If you are the average person wanting to build a really phenomenal Website, most (but not all) of this won't apply to you ... primarily all the nerd stuff about servers and things. This stuff is for the company that hosts your website to worry about and trust me, they already know it. You might want to skip to here.

If, on the other hand, you're thinking about cranking up your own dedicated business grade, (whatever that means) Web server ... this might be a good read.

In any case, Let Me Know

In answer to your question/assumption about “Aren’t Most Websites The Same Underneath …?”

At first blush, Yep, they certainly are!

All you need to see how simple some so-called complicated web pages are is use your browser to look at the HTML source under them. HTML isn't that complicated, even when you add in some basic CSS and simple client-side scripts.

A common characteristic of those who labor under this misconception is that almost all of them had recently created, or at least played with creating a Website. Most had some scripting or programming experience, even if it was just high-school BASIC. Some had even produced quite sophisticated looking and behaving Websites using Microsoft Front Page or a similar automated tool. Most had at least a passing comfort with the HTML.

What's the big deal? Some cutting and pasting, a little formatting and click the "Publish" button. BINGO -- A Website!

ABSOLUTELY TRUE!

With today's automated tools, template based sites and on-line 'Build a Site' resources, the average person with moderate computer skills can do almost anything they need. ( the almost can jump up and get you though)

Of course, buying a word processor doesn't make a person a writer, but I'm not talking about the ugly, inaccurate or cumbersome factor,  just what's going on 'under the hood.'

If that HTML source you are looking at is a 'hand-edited' document, unchanged by the server, you are more than likely correct about the simplicity of the site. The page you are looking at right now for example, is a very simple HTML Document with a bit of scripting tossed in for the flyout menus.

Looking at the source for a Google search result in this way could lead one to believe -- what's the big deal, Google is a reasonably simple Web Page? You'd never realize that there are literally thousands of interconnected servers working behind the scenes to deliver the simple HTML page beneath your Google search result.

Yes, there are quite a few 'common factors' among most Websites. For the Web, the underlying hardware, is most likely an x86 platform, but might also be a Macintosh, Sun or even a corporate mainframe.

There can be from one to dozens, or even thousands of interconnected servers. A Satellite Imaging Intranet I worked on at a three-letter agency ran on a load-balanced cluster of 254 HP UNIX Servers and this was in 1996. Even a medium sized trade association site that we developed in 1998 required two clustered servers to balance the workload from frequent 5000 or more concurrent member logins, downloads and product searches.

Yes, the operating system is most likely (80%) some flavor of Unix, Linux, Solaris, OSx (a Linux variant) or Microsoft.

Granted also, the majority of Web Servers run under some flavor of Apache Server over a Linux variant, but even among these there are an infinite number of possibilities and configurations. It could, however, be a Microsoft IIS Server churning ASP into HTML, and some experts swear by serving up ASP pages on an Apache Server, running under LINUX. ... but it's most likely ...

This is where ‘most likely,’ bleeds to its unemphatic end.

The web server can run under any number of hardware and feature variations, resulting in most servers running some sort of custom compiled Kernel.

Unlike Windows, the LINUX system Kernel can be custom compiled from the source code as needed to incorporate additional drivers or options at the Kernel level, or, more importantly, remove unneeded features and services to gain maximum effieciency from an opearting system custom made, just for you. The same applies to the core Apache Web Server, Mail Servers, SQL Servers and just about everything else. Sites operating at peak efficiency do so because they build and configure the core server's services to fit their unique environment.

Years ago, if you wanted a Website there was only the HTML with which to work. Today there are a variety of Document Display Languages with a generous selection of client and server side scripting languages. Some sites use only client side JavaScript or vbScript to add basic interactivity but most take advantage of at least some server side processing.

The server-side variations are plentiful, both in compiled binary languages such as C++ or compiled BASIC flavors and scripting languages like PERL, served to the visitor via the Common Gateway Interface. (CGI) You are also just as likely to find PHP or PYTHON code underneath a typical site and even FORTRAN and COBOL are in current use where appropriate.

There is no ‘common’ database format used with Websites. The range runs from flat text files to PERL/dbi, Postgres, SQL, db2, Access, Oracle, a variety of bTree and ISAM formats and more, each with strengths and weaknesses.

Code written in Macromedia Flash or Director is common and adds to the variety and complexity of server configurations as does rich media content, often requiring the aforementioned custom compiles of the OS and the Applications to serve up at peak efficiency. Web servers can deliver ActiveX Controls in the form of .OCX files to windows clients and Client and Server side java to any platform and XML makes it is possible to finally separate the content from the style.

Most Websites have an E-Mail component, meaning that there is also an SMTP, IMAP and/or POP3 server to configure (again with numerous options) and maintain. There is also an FTP server or two and often public FTP folders. Two DNS servers are associated with every website (a primary and secondary) and maintenance of the Domain Tables, while highly automated, still has to be done.

There is generally a large and individually configured security component to every Website, starting at the firewall and leading to various user authentication methods and encryption technologies. Linux and Unix systems support rich file, folder, user and group based authentication.

Add to this one or more database servers, dozens of logs and configuration files and an OS that is friendlier at the command line than a mouse pointer and no, not all Websites are alike.

Web browsers are also far from standardized and/or standards compliant, though they are improving. Every document created for public consumption should be tested on multiple browsers on at least Mac (IE, Safari, Netscape Mozilla, Opera) and PC (IE, Netscape, Mozilla, Opera) Platforms.

Subtle differences between the way scripts are interpreted and page elements are rendered between browsers means everything must be developed as the ‘Best Compromise’ when you know that the ‘public’ will be accessing the site. If the buttons don’t line up in IE5 for the Mac the same way they do in Mozilla for the PC you should at least ensure that the page displays legibly and operates properly.

Oops, I almost forgot, visitors coming in from AOL, PDA's and Cell Phones need special tratment.

So, is a Website an IT Project or is it a Publishing Project?

Simply stated, yes!

Like the familiar breath/stomach mint, it's both.

Yes, a Website lives on an underlying Hardware and Software structure, but then again, so does your new car, your DVD player and the thermostat in the hallway.

A Website also shares a significant amount of DNA with more traditional printing and demand document publication methods. The Internet is after all little more or less than an inexpensive, reliable, standardized way to get a whole bunch of wildly different computers to talk to each other. The Internet is to computers what Esparanto only promised to be to language.

A Website is essentially "an interactive, online collection of related and/or linked programs and documents."

One major difference between a Website and many IT projects and/or publications is that a Website need not be ‘finished’ when it is published. In fact, in Internet Publishing, it is poor practice to wait for something to be finished before it's published. (Yes, I know there are exceptions, but before you write me, read about nitty little nerds in the grey box above.)

With minor exception, a Website is never really ‘finished.’ It is much better to publish what you have (as long as it is accurate and works) along with “Bookmark This Site Now! – Coming Next Month, Four New Ways to Sort Your Widget List.”

Repeat visitors are the life’s blood of a site. Rather than losing a Month of exposure for your Widgets to the competition while waiting for everything to be perfect, get it out NOW! Once published, encourage visitors to check back often to see/use your new features.

A Website ‘Grand Opening’ or ‘Total Facelift’ is a one-time, one-shot announcement. “Come See Our Newly Renovated Site!”

Incremental publication and announcements of enhancements are a golden opportunity (and a good excuse) to initiate a fresh visitor contact, send out an announcement or even a press release if the improvement is significant.

Oh yeah, (as Columbo would ask) One Other Thing ... Is Your Website in Compliance with the Americans with Diabilities Act?

While we're on the subject, is your privacy statement up to date, and do you know about the special Child Protection Laws that apply specifically to the Internet?

Some additional guidelines and Web ‘rules’ that may or may not be intuitive are covered in…
The Eggmans (not so) Concise Guidelines for a Successful Website