Meta trend
Through 2003/04, infrastructure consolidation will be driven by value-based portfolio management, but remain impaired by non-linear server pricing, immature tools, service-level priorities, chargeback, and organizational politics. Physical co-location and networked storage consolidation will be widespread during 2002/03. Premium high-end server pricing, coupled with immature partitioning and workload management, will hinder higher-level OS, DBMS, and application server consolidation for Unix (until 2003/04) and Windows (until 2005/06).

[an error occurred while processing this directive]

New midrange systems
During 1H02, HP, IBM, and Sun all announced new midrange to high-end Unix servers. We believe each vendor had different reasons for announcing new server offerings (IBM to fill product line gaps and bring Power4 technology downstream, Sun to close a high-end price gap, and HP to refresh its midrange server technology). But all three new systems are based on their respective vendor's top-of-the-line, high-end offerings and provide many high-end capabilities, rather than being upgrades of low-end boxes.

In addition, the latest round of midrange systems recently announced (e.g., Sun F12K, IBM p670, HP rp8400) is actually comprised of high-end systems based on historical performance levels (150K+ tpm-C). This is evidence of the overall server market dynamic, driven by the steady increase in performance of Intel servers (coupled with more than "good enough" operating systems, such as Windows and Linux). We believe price competition and n-tier application architectures are driving Unix servers deeper into the high end. Intel servers running Windows and Linux dominate at the low-end Web server, network-edge tier, and Windows and Solaris (and, in the future, Linux) at the application tier. This leaves RISC/Unix vendors competing with IBM mainframes for the back-end database server tier.

Although we believe it will be three to five years before Linux platforms will be competitive with high-end enterprise Unix systems for database server applications, by 1H03, Oracle on Linux on Intel ("Lintel") will be a low-cost option for most application workloads. Moreover, by 2H04, we believe next-generation processor technology (e.g., Power5, UltraSPARC 4, HP Intel IPF-64 "Madison Module"), will enable 8-way systems to offer performance levels comparable to current 16-way systems with 2x better price/performance. We believe the trend toward server configurations with smaller numbers of faster individual processors (e.g., IBM with Power, and HP with Itanium processors in the future) will continue until at least 2005/06. As a result, software pricing based on per-CPU pricing models will become increasingly problematic, until Intel and RISC platform partitioning and workload management capabilities mature, and independent software vendors adopt partition-based or resource utilization pricing.

Consolidation: Two for One
For several reasons, we believe two or more systems with strong performance from this emerging category of midrange systems will be better consolidation platforms than one top-of-the-line high-end system, and will mitigate some of the consolidation constraints that make it difficult to justify high-end consolidation. Our research indicates that two midsize systems are 30 to 50 percent less expensive than one very large system. In addition, a two-node cluster will be more reliable than clustering between partitions on a single system, and the midsize option enables more granular upgrades (e.g., to N+1 clusters).

1 2	| Next page

Newsletters
	Tech Update Today eBusiness Update	Tech Update Weekly		All newsletters FAQ Manage my newsletters

[an error occurred while processing this directive]

[an error occurred while processing this directive]