Have we seen the end of Data Centre Evolution?
With the Gartner Data Centre events taking place in London and the United States this week, I cannot help but wonder how the Data Centre has undergone several stages of evolution to command the level of importance that it currently does in any organisation that uses IT. It is essentially the heart and nervous system of any IT system, from where IT services are delivered to business users.
Despite appearances, the data centre industry has an interesting history. In the 1960s, data centre mainframes housed the super-computers (such as the Cray-1) designed for high end computing. Running applications for weather forecasting, molecular analysis, quantum physics and even nuclear simulations were commonplace on these massive machines. These were expensive resources, which only a select few organisations in the country could afford to own and maintain. That changed with the emergence of the micro-computer and better storage technologies in the 1980s, and data centres transformed into server rooms where specialised applications ran.
The real popularity and need for data centres started booming in the dot-com era of the 1990s, when organisations began to require high speed computing power and internet connectivity to host applications on. Due to prohibitive hardware costs, it was practical to rent out such equipment rather than invest capital in it. This led to the emergence of the Web-hosting providers and Internet Service Providers, who could offer such services on a dedicated or co-location model.
Fast forward to today, modern day data centres, responsible for offering applications and platform as a service, have to be designed for massive data volumes at speeds and latencies beyond traditional architectures. Today’s web applications for business, social networks, and mobile devices demand it. A few emerging technologies such as Data Centre fabric promise to help data centre architects meet these objectives by transforming silos of computing and storage resources into a shared and integrated pool of resources and improve application latency.
Technologies such as Virtualization have driven server consolidation in the data centre. An industry peer once told me that they had firm directives from top management to consolidate all their servers from various divisions to the data centre. While the objective was to ‘push’ servers from all-over-the-enterprise to the data center, there was no importance attached to how to model the data centre architecture efficiently to meet this objective. The analogy my associate gave was: “They cleaned up all the mess in the house, but put it in the attic where it’s not visible to outsiders. However, the fact remains that the attic was still in a mess.” What he meant was, the complexity had been moved from all over the place to the data centre, becoming a nightmare for the data centre manager and his ops team. Reasons for this included, heterogeneous network equipment which did not work well together, servers that did not lend well to virtualization, plethora of management tools and a data centre architecture that did not serve applications fast enough to the end user. I completely agree with this view, data centre consolidation is not just about throwing all the servers in a room and virtualizing it - that’s a start but it doesn’t stop there!
Some organisations see data centre consolidation as a tactical way to save costs, but I believe it is a far more strategic goal that can help IT deliver new emerging services in an agile and cost effective manner. If architected correctly, data centres should help reduce power, real-estate, cooling and cabling costs, improve application latency, ensure better resource utilisation via dynamic and flexible resource provisioning. As part of data centre architecting, one of the foremost things to keep in mind is to simplify the data centre architecture and the tools to manage it. Upcoming advances such as Software Defined Datacentres, unified fabric technologies and Integrated IT Management techniques are key to simplifying the data centre and the path to ‘IT Nirvana’.
Providing such agile and flexible resource allocation along with better accounting of utilized resource and chargebacks will enable IT to operate as a profit centre providing efficient and cost effective services. That in all probability will earn IT a seat at the decision making table.
By Sridhar Iyengar, Vice President, Product management, ManageEngine
