Linux network programming

The influence of Linux is beginning to show in corporate technology strategies.

When you bring up Linux, the first thing most people think about is the fact that it's essentially free. Because developers collaborate on the revolutionary OS without being paid--and because Linus Torvalds has remained true to his vision of open-source software--Linux only costs you money if you rely on a vendor's particular distribution. The idea that a modern operating system can be built exclusively through volunteerism is staggering. Yet, it's true.

More importantly, Linux is a great OS. Proven to be more stable than its more widely used, proprietary counterpart, it offers much more scalable performance on commodity Intel-based machines. That's one reason that it's showing up more often as a platform for network infrastructure--i.e. Web, e-mail and domain name services (DNS).

While we consider Linux's remarkable price/performance value proposition for infrastructure requirements, there are two things we shouldn't forget: 1) open source is more than just Linux, and 2) "open source" means more than just freeware.

The first point is important because of the growing amount of software being written and offered according to the open-source model. Of particular interest to Communications News readers is the availability of open-source network and systems management tools. These tools, including increasingly popular Linux-based applications, such as NetSaint and Webmin, can be downloaded free of charge. They offer functions similar to proprietary management applications, such as SNMP monitoring, customizable alert settings and historical reports. What they don't offer is a big price tag or expensive support contracts. Instead, support is delivered through a "community" of users whose responsiveness typically exceeds that of paid services.

THE FUNDAMENTAL DIFFERENCE

The second point is more fundamental. Open-source software, lest we forget, is so named because users get access to the source code of the application. They can freely modify it, add to it and integrate it with other applications. This is different from proprietary applications, where users either have to work with limited application programming interfaces or pray that desired features are built into a subsequent release.

The open-source model thereby gives network managers who possess both the ability and desire to develop their own highly customized management solutions the wherewithal to do so. That's exactly what many of the managers I've interviewed recently have decided to do. They're taking their Unix script-writing skills and applying them to bare-bones open-source management utilities. Within a relatively short amount of time, they're creating management systems that meet their needs simply and cost-efficiently.

Such an approach runs against the conventional wisdom that it's smarter to buy management than to build it. However, a variety of factors are changing the dynamics of build-vs.-buy decisions.

One is the inordinately high cost of management applications. Like their peers in the OS market, management software vendors have priced their wares so high that they made it almost inevitable that the marketplace would somehow respond. Many of these expensive management solutions are also complex to deploy, and offer a wide range of sophisticated features that buyers don't use--either because they don't really need them, or because they're just too complicated to implement. Buyers are thus paying for software functionality they don't need. Open-source packages take the opposite approach, allowing them to start with a set of core functions and add to them as required--at their own expense.

WHAT IT'S WORTH

Build-vs.-buy decisions aren't just based on economics. They're also based on strategic need. The general principle is that if something isn't strategic, it's okay to buy it. If it is strategic, then you build it to ensure competitive differentiation. The willingness of today's network managers to build their own networks can therefore be viewed as a recognition of something I've been saying for years: if network technology is truly strategic, then network management is a strategically critical discipline.

The proliferation of open-source management tools--and the increasing propensity of network managers to embrace those tools--indicates that more and more corporate technology teams are coming to grips with this truth. There's certainly nothing wrong with writing a check and buying a management software package, or even buying network management as an outsourced service, but don't kid yourself. If you go the "buy" route, you're still going to have to do a lot of work to get your bought package or service to fit with your particular infrastructure and service-level requirements. You won't gain any strategic advantage in terms of performance, responsiveness to change or differentiated capabilities. Packaged software and packaged services don't offer that.

The open-source movement has serious implications for network management. It offers organizations the ability to differentiate themselves based on the skill with which they can build highly customized management solutions. This is good news if you have these skills. If you don't, then you can still depend on the cash. But in today's sobered economy, you might want to recount that cash before you decide exactly how you're going to manage your network.

To comment on this month's column, contact me at 11@exit109.com

COPYRIGHT 2001 Nelson Publishing
COPYRIGHT 2001 Gale Group