This article originally appeared at:

Linux's Value Proposition: Integration
By Eric Kidd - Linux
March 16, 1999 12:43 PM ET

Ed Muth, a product group manager at Microsoft, criticized Linux in a recent ZDNet article. As a professional Linux developer, I'd like to look at the issues he raised. In particular, Muth felt that Linux had a weak "value proposition" and a low level of integration among applications.

Companies including IBM, Oracle, Compaq, SAP, Computer Associates, and others have announced support for Linux. Hewlett-Packard and SGI have even established new divisions to work on open source software. So why is Linux so hot all of sudden? Have all the blue chip companies gone crazy? Are they doing this to annoy Microsoft, or do they see Linux as a major player?

Why Linux?
Linux is an excellent Internet server: it's cheap, it's reliable, and it scales well. The most important Internet software runs on Linux: Apache, which serves more than half the Web sites in the world; sendmail, which moves 80 percent of the e-mail; and Bind, which translates computer names into Internet addresses. Linux also runs Perl, the most popular language for creating dynamic Web pages, and several major application servers including Zope and Cold Fusion.

Linux servers have low hardware and software costs up front. Ongoing maintenance is roughly equivalent to that of SCO or Solaris machines -- you'll pay slightly more for system administrators, but you'll save money because of decreased downtime.

Linux 2.2 scales very well to four processors. If this proves inadequate for your business, you can develop applications under Linux and deploy them on high-end Unix mainframes. Sun's Enterprise 10000, in particular, provides excellent Linux compatibility and support for up to 64 processors and 64 gigabytes of RAM. Thanks to Interix, you can even deploy your Linux applications on Windows NT.

But What About Integration?
Integration is a two-edged sword. On the one hand, highly integrated software allows you to combine applications in powerful ways. On the other hand, integrated software is much harder to debug. In some cases, integration can even be used to limit consumer choice and drive competitors out of the market.

Linux developers try to strike a balance between too much and too little integration. All the important server tools can work together, of course. I recently built a financial application that uses Perl to parse stock market data. This data is then imported into a PostgreSQL database and served to my client's intranet using Zope and Apache.

On the other hand, Linux software isn't overly integrated. In the above example, I could replace the database with Oracle and everything would still work. If my client discovers a bug, I can isolate each part of the system from the others and debug each part separately.

So while integration is important, some kinds of integration are better than others. It's best when applications from different vendors can be mixed and matched. The Linux community in particular loves this freedom too much to lock users into a single solution.

About the author: At his previous job, Eric Kidd managed several Linux servers for a small engineering company. Today, he's helping a site with 10,000 users test an important server application under Linux. The initial results look quite promising, thanks to the speed of the Linux ext2 file system.

Interested in writing a Linux FUD-Buster? Please read our Editor's Note explaining our FUD-Buster policy.