Decision Time: Windows Server 2003 Upgrade or Not?
Windows Server 2003 — to upgrade or not to upgrade? That is indeed the question for today's Windows NT- or Win2K-based IT department. In fact, it's a pretty good question for any IT department to be asking right now. As with many seemingly simple questions, the path to the answer can be complex and can require a great deal of consideration. An "Executive Overview" of some key areas of concern can help shed some light on that path.
The question involves numerous key issues, including cost of acquisition and implementation, as well as total cost of ownership (TCO); performance considerations including security, reliability, availability, and scalability; and management and operations considerations. It's a pretty safe bet that no IT department is going to be standing still for too long, so the question is not so much whether or not to upgrade as it is whether or not Windows Server 2003 is the upgrade path to choose. Since the principal visible competitors are the various flavors of *nix servers with Linux leading the field, that will be the primary comparate.
If cost of acquisition marks the first battle, then Linux probably wins the initial round quite handily. It is perceived as essentially free, which is a tough price to beat. Granted, there may be some costs associated with acquiring a particular distribution, but Linux is not typically burdened with the licensing costs that go along with Windows.
According to studies, however, initial licensing costs are the only saving grace for Linux in terms of overall cost advantage, as it quickly becomes more expensive than Windows. Cost of staffing surfaces as the main difference — Windows simply requires less management staff, and that staff generally requires less training.
Analyst firm IDC compared the TCO of Windows 2000 to Linux and concluded that when it comes to network infrastructure, print serving, file serving, and security applications, "the cost advantages of Windows are significant: 11%-22% less over a five year period." Linux did show a 6% saving in the area of web serving; however, Microsoft says they are "delivering more to web server customers by reducing resource requirements and lowering acquisition costs with the release of Windows Server 2003."
Giga Research, a subsidiary of Forrester Research, ran an objective comparison and analysis of portal application development and deployment for .NET on Microsoft Windows and J2EE on Linux. Their report describes a "Total Economic Impact" (TEI) in which they conclude, "Microsoft offers a substantial cost advantage over J2EE/Linux as a development platform for the applications considered."
The cost savings occur largely as a result of Microsoft's tools that "simplify development," which then translates into "lower labor costs for development and administration of custom applications and a faster time to deployment" — twenty-five percent lower in some of their sample cases. Again, it would seem that the savings in licensing costs are more than replaced by higher subsequent labor costs.
When it comes to performance, no guessing is required. There have been numerous independent studies and tests on which we can base decisions. One such test, designed by Veritest and audited by Meta Group, studied IBM's claims about running Linux on their mainframes to consolidate Windows servers, especially for file serving and web serving.
The test used SuSE Enterprise Linux 8.0 as the operating system, the VeriTest NetBench 7.03 benchmark against Samba for file serving, and the WebBench 4.1 benchmark against Apache for web serving. The results demonstrated that "mainframe Linux performed poorly on standard file-sharing and web-serving benchmarks — between 20 and 300 percent below that of Windows Server 2003 on the VeriTest study."
Also, according to Martin Taylor, Microsoft's General Manager, Platform Strategy, "Windows NT 4.0, a product that was released seven years ago, produces NetBench file-sharing results 20 percent below those produced by mainframe Linux. Windows Server 2003 performed over 150 percent better than Windows NT 4.0 on the same hardware, which we think attests to the advances that have been made in Microsoft server technology over that period." This would indicate that even the smallest of shops could get a performance improvement by simply upgrading their OS and breathing new life into their servers.
Consolidating NT 4.0 servers on a Windows Server 2003 platform also utilizes the enterprise's investment in Windows expertise, requiring less retraining than does a change to a Linux platform. This particular point is a recurring theme in the various reports I have referenced here.
Another Veritest study compared the web serving performance on a variety of HP Proliant server hardware configurations between Windows Server 2003, Enterprise Edition; Red Hat Linux Advanced Server 2.1; and Red Hat Linux 8.0. The results of this study showed Windows delivered between 51 and 300 percent better performance — more than enough to offset the six percent web serving advantage for Linux from the IDC study.
The familiar GUI and a comprehensive set of wizards make setting up server roles a simple task. Similarly, routine management tasks are easily accomplished. Active Directory enhancements provide new flexibility to allow AD to change as your organization changes. Transitioning from NT 4.0 domains is also simplified with new upgrade wizards. This ease of transition and management is a large factor in the lower Total Cost of Ownership (TCO) that the analysts are reporting.
Microsoft has poured $200 million into improving security in Windows 2003, and the result is their most secure operating system ever. An interesting observation is that while the various worms and exploits that have attacked the Windows family of operating systems have garnered a great deal of publicity (largely due to Windows being by far the most common OS, especially in private homes), Linux, it turns out, is the victim of more successful attacks (see this Security Spotlight article for the full story). First glance clearly doesn't tell the whole story!
Other noteworthy features include the Microsoft .Net Framework that, as an integral part of the Windows 2003 operating system, provides the basis for high-performance web applications. The shadow copy/restore feature provides a mechanism for users to easily recover their own files from previous versions without the intervention of IT staff. Scalability is provided through support from single-processor systems all the way up to 64-way systems. High-availability demands can be met through the eight-way clustering in Windows 2003 Enterprise and Datacenter editions.
With this latest generation of the Windows Server family, Microsoft has demonstrated its commitment to providing a secure, available, and scalable high-performance operating platform that is (relatively) simple to implement and manage. Detailed studies have shown the true TCO/TEI advantages, so since we know that IT is not going to be standing still, the question becomes one of whether or not the enterprise could (or should) afford an upgrade path based on *nix.
The following links provide access to reports referenced in this article:
Microsoft Windows Server 2003 with Internet Information Services (IIS) 6.0 vs. Linux Competitive web Server Performance Comparison (pdf)
- Microsoft Windows Server 2003 vs. Linux
Competitive File Server Performance
- The Total Economic Impact? of
Developing and Deploying
Applications on Microsoft and
J2EE/Linux Platforms (pdf)
- Windows 2000 Versus
Linux in Enterprise
- Windows Server 2003: Mainframe Linux Benchmark Project (pdf)
This feature courtesy of EITPlanet.