A Second Look at Linux

Last fall I ran my most popular ENT column, gauging from the mail received. Of course popular does not necessarily mean well-liked. The topic was Linux, and my conclusion then was that Linux would remain a niche operating system for the foreseeable future -- perhaps forever.

I responded to the torrent of letters with a promise to revisit my predictions in mid-1999. Here is my promise fulfilled.

There is an active and growing, if not vibrant, Linux community today. The operating system has its own conference and expo, which in the computer industry is an emblem of achievement. Unix used to have expos, too. Several in fact, but not any more.

Furthermore, there have been many product announcements and articulations of Linux strategies coming from some venerable companies.

Despite these and other developments, I still say Linux remains a niche operating system and will remain so for the foreseeable future -- perhaps forever. That much hasn’t changed, only we now have more evidence that this statement is true.

But first, let’s look at the positive side of the Linux balance sheet. As of this writing, IBM was getting ready to ship new versions of DB2 for Linux. IBM had distributed more than 30,000 free test copies of DB2 Universal Database while it prepared to ship a commercial version of DB2 for Linux.

Bigger Linux news comes from Hewlett-Packard. HP is preparing an early summer launch of its OpenView network management system for Linux. Up to this point, HP’s Linux support has largely consisted of rhetoric. OpenView, however, is an HP enterprise crown jewel.

Sun Microsystems, the penultimate Microsoft detractor, recently unveiled a key piece of software that, posing as a translator, allows Linux applications to run on Solaris.

Wait, there’s more. On the database front, both Sybase and Informix are littering the developer landscape with test versions of high-end data bases for Linux. And the big dog of databases, Oracle, has started shipping Oracle 8i for Linux with support for four-way multiprocessing servers.

Then there is an intriguing bit of software from Palo Alto, Calif.-based VMware. Its namesake product is expected to help users run an additional operating system, such as Windows, in a Linux virtual environment. In theory, VMware would allow developers to work on two different platforms concurrently. Early tests of the software show it to be painfully slow, which to developers translates into unusable. But who knows what tomorrow may bring.

So why my pessimism about Linux’s future? Simply because the user community doesn’t share the vendor’s lust for a Windows NT killer. Don’t believe me. Ask the users.

That’s what the International Oracle Users Group of America did when it polled nearly 900 companies migrating to Oracle 8 or Oracle 8i. A not-so whopping 2 percent of these companies plan on deploying the database on the Linux platform. This speaks volumes in light of the ongoing massive giveaways by database companies of developer editions of their products for Linux.

In May, Evans Marketing Research reported results from a poll of 500 Windows developers. One in five of those polled felt Linux is ready for mission critical applications.

In a major head-to-head lab test conducted by PC Week with NetWare, Unix and NT, Linux scored below average on nuts-and-bolts file/print/Web services. Its applications did not offer the scalability required for today’s business-critical environments. It also scored miserably in overall application support and RAID support, two essential enterprise attributes.

Head-to-head tests between Linux and NT presented in NetworkWorld showed similar Linux shortcomings, including weaker hardware and device support than NT and a lack of desktop policy enforcement components.

It’s not just inferior attributes that will continue to define Linux as a niche player. With Y2K issues coming to a head and then spilling over into 2000 with clean-up work, the last thing most enterprise IT people are going to undertake is a major platform switch. Not unless the reasons for doing so are compelling, and that is not the case with Linux.

Thus, I repeat my conclusion of six months ago. Consider putting Linux up on a server in a test cell with Windows-based workstations attached to it. It just doesn’t make sense at this early juncture to make major commitments to Linux, free or not. --Bill Laberis is president of Bill Laberis Associates Inc. (Holliston, Mass.) and former editor-in-chief of Computerworld. Contact him at bill@laberis.com.