Today, as then, if Linux is being snubbed by an enterprise, it's likely for cultural reasons. IT managers often equate open source with shareware and a lack of commercial support. Unless a manager has been involved in a successful Linux deployment, caution is the name of the game. Linux advocates at the bottom of the chain of command face an uphill battle to change this closed mind-set. The status quo, whether it's the reliability of Sun Microsystems Solaris hardware or an unwillingness to evaluate a system that is not based on Microsoft Windows NT/2000, is difficult to change -- even with the promise of increased ROI (return on investment), better performance and improved uptime. It's hard to teach an old dog new tricks. Another factor hurting Linux is market fragmentation -- there are upward of 150 distributions out there, though only a small fraction can be considered enterprise class (we tell you which ones fill that bill in "SuSE Queues Up for a Clean Sweep").
But the pace of change is about to accelerate. In the months that have passed since we made our Neohapsis lab in Chicago an all-Linux all-the-time venue, millions of dollars have been burned cleaning up the viruses and worms that have decimated networks running Microsoft software. Then, on Sept. 19, in the wake of Code Red and Nimda, Gartner Group issued a controversial statement advising that enterprises "immediately investigate alternatives to Microsoft IIS [Internet Information Server]." Given Linux's reputation as a highly securable OS, it's no wonder corporations are beginning to re-evaluate how Linux might fit into their organizations.
What Do Readers Think?
Check out our e-poll results
Even before Nimda hit, Gartner predicted that the Linux server market would more than double within four years, from $1.8 billion this year to $3.8 billion by 2005. We think that estimate may be conservative given shrinking budgets, burgeoning security concerns and the outpouring of commercial support for Linux by major players. Familiar names such as IBM Global Services and Red Hat will lead the way in convincing those who are shy of open source that a Linux implementation can be backed up by a real, enterprise-class support contract (click here for a sampling of support offerings).
Recipe for a Linux Application
Of course, all the support in the world won't help if your applications don't run. Enterprise-ready applications ported to the Linux platform have been available for the past couple of years, and major deployments are following. IBM, Lotus Development Corp., Oracle Corp. and SAP are a few of the most prominent software vendors to offer their wares on the Linux platform (see "Big Blue Does Business With the Penguin").
Deploying one of these applications on Linux makes sense for a number of reasons:
But the feasibility of rolling out your favorite application on Linux depends on a variety of factors. All the following must be examined before you can expect to have a successful Linux-powered rollout:
- Ability to use the hardware vendor of your choice.
- Extremely efficient use of hardware resources.
- Linux's reputation for reliability and uptime.
- Nearly free licensing of the operating system.
Kernel 2.4: Ready for Business
- If the application is not native to Linux, has the port reached a level of maturity comparable with that of its native environment?
- Is a commercial support structure in place?
- Do you have in-house Linux expertise, or can you afford outside help?
- Do you have management-level buy-in for the project?
The 2.4 kernel, a much anticipated release, was first blessed by Linus Torvalds (the "inventor" of Linux and current lead architect of the kernel) in January. Since then, most of its bugs have been worked out, and it has been adopted by every distribution on the market because of its increased performance and hardware support.
By far, the most popular hardware platform for Linux is still Intel, but with 2.4, four new architectures are given native Linux support: IA64 (Itanium), S/390, 64-bit MIPS and SuperH (Windows CE platform). Hardware scalability was a key goal of the kernel developers, and the 2.4 kernel supports 64 GB of RAM (on Intel) plus as many as 16 NICs and 10 IDE controllers.
Talk to anyone familiar with the Linux kernel lineage, and he or she will rave about the increased performance granted by 2.4. In addition, an improved scheduler, memory manager and VFS (virtual file system) all contribute to and are tied into the full-featured resource management subsystem making its debut in the 2.4 kernel. Other enterprise-level improvements include lifting of the 2-GB file limitation, support for more simultaneous processes per CPU and support for 2.4 billion -- yes, billion -- users and groups.
Even with these amenities, the 2.4 kernel remains true to its roots as a resource-friendly OS, not requiring much more memory than its 2.2 forbearer -- in fact, it actually requires less memory than 2.2 does in some situations.
Add all this up, and it becomes clear that Linux will give you the most bang for your buck in situations where deployment of a given application is feasible. However, beware of blind cheerleading of Linux by those who have their own agendas at heart. Linux is not a silver bullet; like any technology, it must be evaluated objectively and applied only where appropriate.