META Trend: Windows 2000 Datacenter enhancements, combined with Microsoft Corp's relentless pursuit of IT/ISV infrastructure developers, will enable more agile NT applications to overtake the functionality of Unix alternatives by 2001/02. However, while NT scalability will be adequate for 90%+ of all application requirements on IA-64 (2001/02), Unix will continue to provide more robust enterprise-level capabilities (partitioning, workload management, clustering) through 2003/04.
By 2004, Microsoft Windows 2000 (W2K)-based architectures will be the most prevalent operating system at each tier of the n-tier model. The W2K family will be capable of supporting more than 90% of all n-tier application and database workload requirements. Through 2002, a mixed mode (operating system - e.g., Unix/W2K, S390/NT) will be the most typical pattern in e-business deployments. However, this mixed strategy will be a key step to an all-W2K/Intel n-tier deployment. In end-user deployments, 70% of all mixed-mode deployments will transition to all W2K. In application service providers (ASPs), Solaris will remain strong, equaling Microsoft's W2K server shipments. Thus, it is critical for clients to manage the transition to W2K and plan for an adaptive Intel Corp.-based infrastructure that will enable different infrastructure tiers to test, roll forward (roll back if necessary), tune, and roll out into production an all-W2K strategy. Use of switch-based architectures and storage-area networks (SANs) is key to enabling the maximum flexibility required for this adaptive infrastructure transition. META Groupresearch indicates W2K is more robust and mature than NT, and an NT-to-W2K upgrade strategy is now viable and should be accelerated.
Through 2001, the traditional Microsoft release schedule of service packs will become less popular in all Microsoft products. As Microsoft's ".Net" strategy is delivered and W2K Datacenter Server (DCS) is licensed, distributed, managed, and supported by the DCS OEMs, the traditional high costs of Microsoft upgrades will be reduced. As the whole W2K suite of products matures, the total cost of ownership (TCO) will be more in line with those of Unix. Therefore, when clients are predicting or using TCO comparisons of different systems, it is key that the TCO infrastructure models are of the same model type and not diversely different model types (e.g., a two-tier or widely distributed unmanaged NT system is not a fair comparison to a tightly managed, consolidated n-tier Unix system).
- The 12 Disciples of Microsoft's Datacenter:Microsoft has selected 12 key OEMs for its W2K DCS partners: Amdahl Corp., Bull, Compaq Computer Corp., Dell Computer Corp., Fujitsu Siemens Computers, Hitachi, Ltd., Hewlett Packard Company, International Computers Limited, IBM Corp., NEC USA, Inc., Stratus Computer, and Unisys. The vendors are varied, holding different values for technical innovation, partnering/support, geographic coverage, and specialist/niche and datacenter presence. As a result, the vendors will fall into a tiered ranking, a broadly commercial Tier 1, comprised of Compaq, Dell, HP, IBM, and Tier 2, comprised of the rest. Niche is not necessarily bad, but the focus required to be successful in a certain niche (e.g., geographic or technology) is contrary to that of broader commercial success.
- Microsoft's W2K Family: A Staggered Transition: Microsoft's W2K family of products (Server, Advanced Server, DCS) fit neatly to each tier of an n-tier strategy (Web, application, and database). The Tier 1 and Tier 2 W2K vendors can be mapped onto each tier accordingly. The Tier 1 vendors have success at every tier; therefore, technology and funding can be spread and can benefit across products from Tier 1 vendors in all tiers. This helps differentiate Tier 1 from Tier 2 and the positioning of the vendors. The different tiers of the n-tier model require differing levels of hardware features and robustness, high availability, scalability, and load balancing. Therefore, each W2K product requires a different rate of adoption into our clients' n-tier infrastructure models:
- Database transition to Windows 2000 Datacenter Server: The increased scalability of 8-16-32 processors for W2K DCS means increased competitiveness in the database tier. The integration and services should be sought from the W2K DCS OEMs (TCO increases will reflect this integration). Unix will be the main competitor to W2K DCS at the database layer. Existing 4-way NT servers should be upgraded to W2K Advanced Server for database use.
- Application transition to Windows 2000 Advanced Server: The better scalability and robustness of W2K Advanced Server (AS) over NT means that clients should move to W2K AS for ERP applications and general and custom applications, after integrating, testing, and optimizing W2K into existing infrastructure. New application deployments should be rolled out on W2K AS rather than NT. Solaris will be the main competitor at the application layer. Linux's main challenge is to become established in the application layer.
- Web transition to Windows 2000 Server: Clients should move to W2K server from NT as soon as possible. All new Microsoft Webserver deployments should be on W2K. Again, the extra robustness and memory/application management offered by W2K Server over NT should be used. Linux and Solaris are the main competitors for Microsoft at this tier level (all three are viable).
The database server market is the main challenge for Microsoft. For a start, Microsoft does not own the database layer on W2K/NT; this is a crown shared with Oracle (around 30%-40% market share each). In addition, SQL Server 7 and NT both did not scale well over 4-8 processors. Although the NT model enabled numerous deployments, the system images were fairly small. As a result, the Microsoft model was tarnished as one of "many distributed servers." This gave Unix a comfort zone for high-end SMP (4-64 processors) and better robustness, scalability, partitioning, and workload management. With SQL 2000 and W2K DCS, Microsoft can address more back-end business than before. Unix will thrive in this tier and the Unix market value will be maintained as system images increase in size, even though Unix market volume will diminish. Multi-partitioned platforms require multiple operating systems; this means capturing and forecasting the OS market becomes more difficult for the statistical research organizations (as hybrid systems become difficult to define).
Business Impact: Failing to move established Microsoft deployments from Windows NT to Windows 2000 will drive higher operating costs.
Bottom Line: Users should review NT deployment and move forward with a NT-to-Windows 2000 upgrade as soon as possible. Users need to align the Windows 2000 server to the matching tier in an e-business n-tier platform. This will enable applications to be more robust and help address some key technology inhibitors present in NT that are overcome with Windows 2000.
Copyright )2000 META Group Inc. SERVER INFRASTRUCTURE STRATEGIES is published by META Group Inc., 208 Harbor Drive, P.O. Box 1200061, Stamford, CT 06912-0061. Web: http://www.metagroup.com. Telephone: (203) 973-6700. Fax: (203) 359-8066. This publication may not be reproduced in any form or by any electronic or mechanical means, including information storage and retrieval systems, without prior written permission. All rights reserved. Reprints are available.
SIS 17 November 2000.873