Essential server strategies for Internet success

The Internet is creating a shift in the focus of business computing. Companies can no longer afford to focus primarily on their back-end systems and internal computing requirements. Instead, they must redirect significant resources to establishing and enhancing their entire Internet infrastructure and customer-facing applications. Craig Brunsden, Intel manager at AXiZ, discusses the effect these changes are having on business strategies.

"Traditional data centres provide businesses with a contained and predictable computing environment," says Brunsden. "Capabilities can be added in a relatively planned fashion. Before the Internet, all business-critical computing occurred in the data centre, where virtually no expense was spared to ensure high performance and availability."

However, the Internet emerged based on different principles, ensuring a high level of fault resilience through distributed and redundant systems and services. Today, these two computing models are colliding as IT organisations scramble to link their data centres to the Internet. Many are faltering. High-end systems that were built for bulletproof reliability are being stressed to the point of failure by new e-Business requirements.

"A solution to this problem can be found in the Internet model itself," says Brunsden. "Companies can `scale out` using multiple affordable servers configured to provide server-level redundancy throughout their business computing infrastructure. This gives IT organisations a clear choice when building solutions - they can apply either the "larger system is better" (scale up) model or the redundant system (scale out) model."

The scale out strategy has significant advantages. First, it`s highly cost-effective to deploy volume servers instead of proprietary systems. Computing resources can be expanded in a modular fashion to more precisely match e-Business growth and to meet changing demands caused by the ongoing Internet revolution. New applications and services can also be deployed more quickly, and workloads are easily balanced across multiple systems to improve performance and availability. This incremental approach also helps businesses take more rapid advantage of the technological innovations that are continuously arising in the horizontal server marketplace.

Once an appropriate level of redundancy has been established by scaling out, a scale up strategy can be used, deploying larger systems to meet increasing demands with fewer servers. By scaling out and scaling up on a consistent architecture, businesses can benefit from the economics of volume, while quickly deploying the most cost-effective and efficient solutions available.

As more critical systems are deployed outside the data centre, customer-facing systems become the new epicentre of e-Business. It`s therefore essential that IT organisations incorporate data centre best practices across their entire computing environment. This gives them the best of both deployment models. Deploying redundant systems based on the inherent availability model of the Internet, and the proven management techniques of the data centre combine to deliver robust and highly available systems.

"In developing an Internet strategy, it`s useful to revisit the concept of availability," says Brunsden. "Traditional data centre systems ensured high application availability within their contained environment. However, Internet users measure success in terms of overall Internet service availability, which depends on the entire computing infrastructure - from linked back-end applications all the way to the Internet connection."

Deploying redundant systems in server farms helps meet the need for Internet service availability by eliminating single-points-of-failure throughout the server infrastructure. The use of cost-effective volume systems means that individual servers can become field replaceable units (FRUs). This approach minimises the impact on service availability during fault recovery, failure repair and system upgrades.

As the Internet has evolved, the value to users has increased and user expectations have grown. In the first generation of Internet business, customers could view static Web pages. The 2nd generation added e-Commerce functionality.

"Intel and other Intel Architecture (IA) industry leaders are currently working on the 3rd generation of Internet technologies," says Brunsden. "These new technologies will introduce a customer-centric strategy, with systems and applications that businesses will use to enhance their customers` success as much as their own."

These new applications will deliver the next level of value to customers by integrating key business processes with customer-side business systems and personal productivity applications. However, even as many IT organisations are struggling to deploy second generation solutions, users are already complaining about information overload and the difficulty of integrating information from multiple sources. To address these issues, Intel and the horizontal IA industry are moving quickly to enable future solutions that make it easier to deal with Internet growth and third generation e-Business requirements.

The IA-64 platform, currently under development, will have the capacity to handle the largest industry workloads on both standard and large-scale system architectures. IA-64 will bring the economics of volume, freedom of choice, and the value of vendor innovation to back-end applications that have traditionally been deployed on mainframes and proprietary RISC-based systems. In addition, the InfiniBand* architecture will bring new I/O capabilities that will further extend the ability to "scale out." It will offer an even more modular approach to system resources, enabling companies to deploy the optimal mix of processing, I/O and storage capabilities to meet evolving needs.