In the past year, the perception of Linux as a serious operating system for people other than "do-it-you-selfers" and programmers has gone from invisible to in-your-face. Linux is popping up everywhere, like mushrooms after a rainstorm. Every day there are announcements of new initiatives that involve Linux in some high-profile application, whether it's Linux-based supercomputers used by oil exploration companies or hand-held Linux PDAs or the GNOME foundation's promise of delivering a common desktop environment that can actually provide some useful services and applications across all Unix platforms (unlike the failed OSF CDE effort of the early 1990s).
Despite all the glitzy announcements and high expectations, there is still a lot of hard, serious groundwork to be done to ensure that Linux systems are successful players in the enterprise computing arena.
For the next several Linux in the Enterprise columns, I'll explore some key services and facilities that are needed to make sure that Linux deployments are successful and will allow your systems to be integrated as seamlessly as possible into your firm's existing infrastructure. These tools and facilities will also be useful to you if your definition of "enterprise" is a small company or start-up that isn't in the Fortune 500 just yet. Among the areas we will consider are: infrastructure monitoring, backup and recovery, automated installations, enterprise file sharing, security and automated log analysis.
One of the most important areas that you will need to tackle if you are planning on deploying Linux systems in an enterprise environment is how to integrate the Linux systems into any existing infrastructure monitoring tools.
At first blush, this might seem like an easy task -- after all, Linux suppports SNMP (the Simple Network Management Protocol) -- but SNMP alone is, for real-world applications, a way-point on a journey and not a destination unto itself. Even though Linux supports SNMP, tools are still needed to make that capability useful. The ability to look at SNMP variables or walk through SNMP MIB by hand isn't gonna cut it. What most people want in a network monitoring and management tool is something that lets them see a nice graphical representation of their network and other facilities and supports a lot of automated, scripted interfaces to manage the systems with a minumum of effort.
There are several tried and tested systems that can allow Linux to be managed just as easily as other systems.
Most large data centers use tools such as IBM's Tivoli, Computer Associates' UniCenter, or Hewlett-Packard's OpenView to monitor systems and give operations staff control over the systems and applications running on them. Fortunately, if you're inclined to buy such applications rather than build, all of these products are available for Linux systems. Unfortunately, Linux systems are supported as clients that can be monitored rather than as monitoring systems themselves, but half a loaf is better than none.
Tivoli, Unicenter, and OpenView have been around for a long, long time, and are used in most of the world's largest computing environments. They have a great track record and IBM, Computer Associates, and Hewlett-Packard should be applauded for being way ahead of the market in their support for Linux.
With tools such as these (they all support basically the same kinds of functionalities, but differ in their implementation and exact feature sets and add-on components), it is possible to do any and all of the following, without a single human being ever touching a keyboard:
You might be thinking "So what? I can do that with some Bourne shell scripts..." -- that's mostly true, but can you perform these action on 10,000 machines simultaneously? Will your scripts scale to a world-wide enterprise? Probably not ... and that's not a "bad thing" -- that's why enterprise management systems were developed.
Computer Associates' UniCenter
With the addition of enterprise strength management systems, you can put Linux systems into the heart of any data center and be assured that your systems can be monitored and managed as well as more "traditional" systems such as Solaris-based Suns or even mainframe systems such as the IBM 3090.
In the next article in this series, we'll cover some that can bridge the gap between the "several servers" model of management and the kinds of enterprise management represented by the Tivolis, Unicenters, and OpenViews of the world, and look at some open source initiatives that may one day provide yet another option in large scale systems management.
David HM Spector is President & CEO of Really Fast Systems, LLC, an infrastructure consulting and product development company based in New York
Read more Linux in the Enterprise columns.
Discuss this article in the O'Reilly Network Linux Forum.
Return to the Linux DevCenter.
Copyright © 2009 O'Reilly Media, Inc.