Companies are waking up to the fact that increased performance and
utilization can help reduce the cost of operations. They also realize that
getting more work done with fewer servers lowers the cost and complexity of
managing the IT environment, ultimately making more funding available for
desperately-needed services that actually improve business.
However, issues such as large caches, out-of-order execution and speculative
pre-fetches persist. Throughput computing is designed to help resolve the above
divergent trends by providing higher amounts of delivered performance and
computational throughput while simplifying the data center.
In computing infrastructures where application and database life cycles play
a fundamental part in the selection of any server platform, there is a wide
spectrum of choices based on key criteria, beyond the price of hardware.
However, organizations are asking their IT executives to choose between vertical
scaling systems (those with more than eight processors) and horizontal scaling
systems (those with up to four processors). Often, IT executives are pressured
to treat this issue as a binary decision or a black-and-white choice. They pick
one system with little or no consideration for fundamental requirements of the
application or workloads.
It is important for companies to have an environment that has both vertical
scaling systems and horizontal scaling systems. The new commerce model created
by the straight through processing from pervasive computing devices to
large-scale disk storage pools requires the server platform to support a very
fast input/output (I/O) environment. Vertical scaling systems provide the IT
backbone for database and application processing. These workloads typically have
very long lifecycles and therefore need to be placed on platforms that will
sustain their growth or changing function needs. CIOs must be aware of the need
to maintain platform stability through long-term system upgrades that are timely
and non-disruptive.
Unlike choice, the scalability of an operating system is often taken for
granted because it is a non-entity in horizontal scaling systems. With effective
planning and utilization, CIOs can hope to achieve optimum utilization of
resources.
Consolidate and win
Today's business environment is characterized by global opportunities and
challenges. Quick identification of opportunities is critical to growth. The
answer to many of these challenges is "server consolidation". It is a
way to centralize business computing workloads to reduce cost, complexity and
management overheads.
|
Recent studies show that IT costs increased by 220% from 1996 to 2001,
whereas between 1990 and 1995 these same costs increased by only 56% (Source:
ITG). Ongoing recurring costs represent at least 70% of Total Cost of Computing
(TCC) on a five-year basis, and the portion is expected to grow to nearly 90%in
two to three years (Source: Meta Group).
One major cause of this growth in IT costs is the growth in the number of
servers in companies. It is in this context, and within a competitive and
changing environment, that the need for server consolidation has arisen.
Four stages of server consolidation
There are four stages or types of server consolidation that customers are
implementing which offer a wide range of business value and varying degrees of
solution complexity and investment. The first stage is centralization, commonly
called data center consolidation. By relocating existing systems to fewer
numbers of IT sites, economies of scale can provide simplified management and
cost improvements.
The next step to IT optimization is physical consolidation, which consists of
replacing some numbers of smaller servers with fewer, and more powerful systems.
This consolidation, however, is achieved within the same architecture, such as
IBM Unix servers to bigger IBM Unix servers, Intel servers to super Intel
servers, and so on.
|
The third and the fourth type of server consolidation-application
integration and data integration, can gain additional benefits. While these are
more often complex projects, they can provide significant RoI through
elimination of duplicate and inconsistent information, reduction of
architectures and vendors, reduction of number of applications and servers, and
simplification of system management and security.
With contributions from editorial advisors -Kapil Sood, Director,
Telecom, Sun Microsystem, and Jyothi Satyanathan, Country Manager,
eServer pSeries, IBM India
Consolidating for change
Let's
look at some of the common concerns of IT executives today.
- How do I make business data available for analysis?
- How can I better control costs?
- How can I better support and manage my network?
- How can I better integrate distributed applications and data?
- How do I plan for disaster recovery?
- How do I move from a Web presence to e-business, enabling my business
applications and maintaining high availability/security?
The answer to all these questions is: server consolidation. It is an
enabling technology that:
- Optimizes and simplifies existing IT infrastructure
- Integrates existing architectures across applications/data
- Provides a foundation for new solution investments
- Aligns the IT infrastructure with current and future business requirements
and goals