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]

During the next five to six years, Global 2000 enterprises will bifurcate their high-end Unix platform spending. Our research indicates that 60 percent of the high-end Unix server market will continue doubling workloads year over year--the final lucrative market for the high-end vendors (e.g., IBM, HP, Sun). Workload management features traditionally associated with the mainframe (e.g., robust hardware and software partitioning) are now being offered from the Unix platforms. As a result, the physical boundaries between individual systems are softening and becoming less relevant. The service delivered to a user or application is increasingly based on applications that are distributed across multiple servers or are sometimes hosted within a partition on a single server. It is these logical, flexible boundaries between applications that have become important. The resources available to an application can now be defined by management software that allocates system resources such as CPU, memory, and input/output to the application. We expect all high-end Unix vendor offerings to converge to a fully dynamic hardware- and software-based platform by 2006.

Through 2004, the remaining 40 percent of servers, where high-end servers are used only for consolidation, will face aggressive competition from the well-managed commodity servers and eventually lose out to Intel offerings from HP, Dell, and IBM--diluting the market with additional price pressure. Management and policy for this consolidation entails use of tools such as HP/Compaq Insight Manager and VMware.

Application Virtualization. As Web services move business logic into the application framework, servers are becoming virtual spaces in a fabric of networked resources. Each virtual space is an independent computing environment with logically defined security boundaries, fault isolation, and control over resources that are hardware components. This new flexibility will continue to change the way applications are deployed and how IT operations teams manage virtualization of the hardware, abstracting it from the application and operating system. This process takes place in three stages:

Stage 1: Inheriting the data center legacy
The lesson learned from mainframe software partitions or logical partitions (LPARs) was that if an enterprise has application services with varying peaks and valleys of use, it is more efficient to aggregate the applications and share infrastructure resources. Through consolidation in the data center, resources can be reallocated on the fly as needed, instead of having dedicated but infrequently used hardware to support each of these application services. Using this approach, resources can be made available when and where they are required. If this is done correctly, businesses can reduce or manage the total expense of their computing infrastructure through reductions in the cost of support and skills, capital equipment, and system management. IBM has continued this software LPAR approach with its iSeries (AS/400) and, to a lesser extent, its pSeries (RS/6000).

Stage 2: Hard partitions--the data center complement
Hardware partitions enable clients to protect production environments while system updates or revised applications are tested in an isolated workspace that cannot affect the production system. High-end Unix platforms such as those from Sun and Sequent introduced hardware-based server domains (of which Sun claims to have shipped more than 7,500 of its UE10000 platforms). Each domain runs its own operating system and provides fault isolation, so that system or application errors cannot impact applications running in other domains. Domains can be dynamically partitioned into regions as a result of the hardware switch. In the event of a resource shortage in one domain, the system can automatically borrow resources from another domain.

Stage 3: Full dynamic partitions and resources
Unix vendors such as IBM, HP, Sun, Compaq, and Fujitsu Siemens are taking the server virtualization theme one step further by allowing domains to be partitioned to sub-CPU granularity, using software partitions such as IBM z/OS LPARS, HP virtual partitions (VPARs), and Sun Solaris containers in 1Q03, and AIX LPARS in 1Q04. This will provide great flexibility in provisioning sub-CPU application services, because the application environment will be transportable from one server partition to another, with minimal management overhead. Application services will have a robust isolated execution environment wherever they are provisioned, so it will also be easier to consolidate applications onto fewer footprints.

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]