Over the next year, IBM will continue to formally adopt the brainchild of Linus Torvalds. Linux, the Intel-based open source version of UNIX, is growing in the consumer marketplace of servers at an astonishing rate. According to the International Data Corporation (IDC), from 1997 to 1998, the Linux growth rate was more than 200 percent. Over 17 percent of server licenses registered on Intel machines were for Linux in 1998. In that same year, estimates of the number of Linux users were more than 10 million strong. Meanwhile, IBM has initiated formal relationships with four major Linux developers: Caldera Systems (www.calderasystems. com), Red Hat (www.redhat.com), SuSE (www.suse.com), and TurboLinux (www.turbolinux.com). IBM has also released or announced plans to release versions of DB2, Lotus Domino, MQSeries server, and WebSphere for the Linux operating system.
What does it mean to AS/400 sites? Will the past operating systems wars and the battle for corporate resources once again heat up? Will your staff be further subdivided into factions supporting the Linux operating system vs. Windows NT/2000 vs. OS/400? Just when some thought it was safe to make Windows NT/2000 the standard networking platform, Linux threatens to unravel their best-laid plans.
Whats the Story?
IBMs growing support of the Linux operating system is a good news/bad news scenario for AS/400 customers. The good news is that the cost of building non-AS/400 network nodes just went down. The availability and diversity of Linux versions from a variety of vendors, coupled with the low cost of Intel-based servers, will drive the price of purchasing a server to a new low. There is a rapidly expanding market of Linux application servers, too, and IBM is positioning its primary software products to cash in on this new market segment.
The bad news is that the cost of maintaining a non-AS/400-centric user network will probably go up. Most companies that have AS/400s and NT/2000 server networks will be sorely tempted to install one or more Linux servers for specific applications. This may include Domino, MQSeries server, and WebSphere. But this temptation will also create a requirement for more expertise, and that may further fragment the IT staff. Will this lead to a higher cost of maintenance and a lower ROI for the entire IT organization? I believe this is
more than likely. This, of course, isnt a fault of Linux but the result of the diversity of servers in the server farm. And this is not the only blow-back from the rise of Linux.
Within AS/400 shops, Linux is creating a lot of confusion in the server consolidation strategies of AS/400 managers. Medium-sized shops have been successful at integrating Windows NT servers with the AS/400 using the AS/400 Netfinity server. This popular option on the AS/400 allows multiple NT servers to reside under the hood of a single AS/400, centralizing the operations management of the network and reducing the costs associated with a mixed operating system server farm. However, at this writing, IBM has no announced plans to create a similar AS/400 Netfinity Server interface for Linux. The debate within IBM questions the usefulness of such an interface. After all, the AS/400 already has a UNIX-like environment within OS/400, and Rochester continues to encourage UNIX developers to port their server applications directly to the OS/400. Why, IBM wonders, should it spend the resources to create a brand new AS/400 Netfinity API for Linux? This non-Linux argument makes a lot of sense to IBM engineers. After all, its one thing to create a Netfinity interface for a proprietary operating system such as Windows NT/2000. On the other hand, its quite another thing to create such an interface for one or more versions of a free open-source operating system, especially when UNIX-to-OS/400 APIs already exist.
Of course, from the customers perspective, this technical engineering argument ignores a basic economic fact: Its generally cheaper to purchase an application server designed for Linux than to purchase the same application server for OS/400. And this is true even if IBM provides special packaging and configuration options.
Take, for example, the case of Domino for the AS/400 and its latest AS/400 offspring that IBM calls the Dedicated Server for Domino (DSD). IBM originally took some heat in the press for pricing Domino for the AS/400 at a premium. The original price of Domino for the AS/400 made it one of the more expensive implementations on a per- client basis. Sure, it is a highly scalable solution, the complaint acknowledged, but how many AS/400 sites really need a scalability of 15,000 Domino clients?
Domino for LinuxBye-bye, Bumblebee?
IBMs response was to introduce the DSD: a low-priced AS/400 that was solely devoted to handling Lotus Domino. IBM priced the DSD to compete against configurations of Domino running on NT. In that scenario, IBM identified the combined networking infrastructure costs for typical NT-plus-Domino networks as a market opportunity for a new low-cost AS/400 server that ran nothing but Domino for the AS/400. And this strategy worked. The cost-per-user numbers were closely matched between both configurations, and IBM rightly estimated that it could charge a slight premium for the DSDs integrated solution. This has been a relatively successful market strategy for IBM, and the DSD Bumblebee has proven to be moderately successful, with an estimated 16,000 units shipped in the brief time since it was announced.
But now Lotus has introduced a Linux version of Domino that changes IBMs price-point target. This new Linux version of Domino lowers the entry cost for the entire Domino marketplace. How? Even if the Linux version of Domino is priced at exactly the same price as the Windows NT version, the overall purchase price of a Linux implementation of Domino will be lower. Why? First, the price of the Linux operating system is 300 percent less expensive than Windows NT. As a result, the value of the AS/400 DSD in the marketplace has suddenly changed. Instead of the DSD being cost- competitive with the NT solution, the DSD may be the high-priced spread in the low-end Domino market. Second, the value of the entire DSD solution may look suspicious to traditional AS/400 customers. Even though these customers may love the AS/400, would they be willing to pay a premium for a DSD that looks to be no more than a crippled model 170? Are the engineered values of the AS/400scalability, reliability,
etc.powerful enough to warrant the purchase of a machine that can run nothing except Domino?
The same economic dynamic will be at play for every application server that IBM ports to Linux: It will increase downward pressure on AS/400 solutions to be more price competitive. It may make AS/400 customers think twice before they consider another AS/400 solution.
Which Side Are You On?
So how should AS/400 IT departments view Linux? Is it an opportunity or a curse? Should a manager consider Linux to be a serious solution for the IT arsenal? The answer is, Of course! Linux is a cost-efficient and stable solution that stands to win more and more vendorsincluding IBMto its side. But the decision to use Linux in a shop should be viewed in light of the total cost of ownership (TCO) and the need to maintain and support the server after its initial installation. This will continue to be a complex and difficult evaluation. Windows NT succeeded because it was easy to install and the support was readily available. Windows NT looks and feels like other Microsoft products. Linux, by comparison, comes from the UNIX communitya community that thrives on complexity and detail. Its ease of installation and ease of use are significantly different from Windows NT. Versions of Linux come from a variety of sources that may or may not offer a sustainable support profile. Of course, if an organization already has mixed AS/400-UNIX expertise, the implementation of Linux will seem to be a natural extension of IT services and may ultimately remove the need to support Windows NT servers in the shop.
But if the organization is already focusing on a Windows NT/2000-AS/400 consolidation strategy, Linux will be a challenging and disruptive (though exciting) influence on the organization. This influence will grow in the next few years, lowering the financial threshold at which new application servers can be implemented. Indeed, if companies are concerned about the lack of choices available to them, the child of Linus Torvalds may prove to be the most exciting new kid on the block. Regardless of the AS/400 communitys acceptance of this new face in the server gang, it is clear that IBM has chosen to adopt Linux into its family of server solutions.
LATEST COMMENTS
MC Press Online