Data Center Concepts

Planning a data center used to revolve around long-range (e.g. 10-year) planning projections drawn from a server-centric perspective that was enterprise applications driven. That world is largely gone. Certainly, large organizations still rely on enterprise-class applications to drive core business processes. But demand is growing for end-user focused applications designed to help knowledge workers not only process tasks and analyze results, but also to more effectively forecast/predict future needs. These new applications draw upon the increasing volumes of data that organizations collect in the course of operations to fuel their value. Supporting this world is a changing data center environment. Today, a single high-performance server can do the work of multiple older servers. Combine that with the consolidation effect possible through virtualization which is helping drive higher utilization of server resources, and you get an entirely new portrait of data center resources. First, you have fewer servers running at higher density driving more complex cooling dynamics. Next, the relative balance of resources has changed. In the past, you might have planned for multiple racks of servers calling to a few storage devices. Today, that balance is dramatically shifted, with fewer servers calling to increasing numbers of storage devices. Complicating all this is the growing demand for power that many modern technologies demand.

Industry thought leaders are increasingly favoring a data center design-build strategy that is tightly aligned with evolving capacity demands. Using a more modular approach to data center deployment, companies seek to better deploy capital while simultaneous ensuring an ability to meet emerging new demands with rapidly-deployable capacity. This approach also creates greater resistance to obsolescence due to new technology demands such as increased power or cooling efficiency. Modularization also can be conveniently dovetailed with consolidation strategies which are often spread over time due to analysis and transitioning requirements. While modularization may be attractive, it is important to note that effective implementation requires development of an overarching engineering plan that defines best practices for core infrastructure such as power distribution/management, security services and network traffic management.

In the same report referenced earlier, IDC notes that while cloud computing accounts for less than 2% of IT spending today, they forecast nearly 20% of information will be processed via cloud-based resources and as much as 10% might be maintained in a cloud facility by 20151. How to best leverage cloud-based services (public, private, hybrid) is now central to data center planning. Interestingly, the rationale for migrating to a cloud strategy is not always cost or energy efficiency. In Wall Street and Technology, Mordechai Finkelstein, an architecture and engineering executive with Citigroup, noted that Citi's objectives for deploying a private cloud were to gain agility by reducing time to market, deliver improved scalability, and to effectively reclaim unused resources.

A 2011 IDC study titled "Extracting Value from Chaos" forecast 1.8 zettabytes (1.8 trillion gigabytes) of digital data would be created and replicated in 2011, a value doubling every two years1. The study forecast that IT departments would be managing 10X the number of servers (virtual and physical), 50X the volume of data, and 75X the number of files or containers that encapsulate this data. No industry segment is immune from this trend. All this data will be accessed by an increasingly diverse array of applications, including a dramatic increase in demand for access from mobile devices. Plus, beyond primary storage, the growth in data requires unique strategies for the management and backup of this data, including deduping processes and purpose-built backup and redundancy facilities.