Normally, when the topic of datacenters is raised, people start to fall asleep. In the IT world, it has never been viewed as an exciting topic or an exciting market. In fact, this may be the first time that the words ‘datacenter’ and ‘exciting’ have been used in the same sentence. Many of us associate datacenters with soulless grey rooms located in soulless grey industrial estates in the least interesting parts of our towns and cities.
However, this perception is changing as it becomes clear that datacenters are the foundation of cloud computing. Cloud computing is, of course, the most popular IT-related topic, and one of the fastest growing IT markets. The benefits of cloud computing can be clearly explained in non technical language as it has become a strategic issue for businesses, not just an IT-specific issue. Business leaders are now interested in using cloud computing in innovative new ways, such as bringing new products and services to market rapidly, mitigating risk and managing peaks and troughs in demand.
In the same way, the role and importance of datacenters is increasingly being explained in non technical language to a non technical audience. Business leaders are also focused on using datacenters to enable cloud computing.
Why is this? Well, legacy IT architectures and legacy datacenters act as a bottleneck to the development and proliferation of cloud services. Well designed datacenters can enable the optimal delivery of IT resources on demand. In order to meet these new demands, we are witnessing a datacenter renaissance. Massive focus is being placed on the design and future proofing of datacenters as the demands from cloud computing escalate.
When designing a new datacenter or re-designing an older one, it is critical that several datacenter characteristics are considered. The five characteristics that must be evaluated are:
#1 Power is now a bigger issue than space in the datacenter world. Densely packed blade servers consume much higher amounts of power per rack than was the case with earlier datacenter designs. But, less space is required for the same amount of datacenter capability. New datacenters should have the ability to offer 20Kw-30Kw per rack.
#2 Architecture is critical. The datacenter architecture needs to be simplified or ‘flattened’. In many cases, traffic continues to flow through different tiers, which adds to latency with each hop, thereby affecting the performance of real-time applications. In recent years, there has been a significant improvement in the performance of switches, which enables a shift to two-tier architecture by eliminating the aggregation layer. Furthermore, the elimination of a tier reduces the number of switches and cables required in a data center network, leading to lower costs, both operational and capital.
#3 Energy Efficiency. The enormous power demands of today’s datacenters are not only difficult to meet, but also increasingly expensive. The cost of providing power to datacenters is now becoming a very significant line item. It is a cost which can be managed. Datacenters must be designed in a way that maximizes energy efficiency. This may include the use of on-site power generation or/and the use of renewable energy sources. Metrics such as Power Usage Effectiveness (PUE) should be used to maximize energy efficiency.
#4 Location. Latency is directly related to the location of a datacenter. The organizations that place the highest demands on IT require the lowest latency, and consequently they require a datacenter that is located near to the bulk of IT use. Data sovereignty is also a major issue leading to growing pressure for datacenters to be built in the country where most usage is anticipated.
#5 Security. As I have mentioned in earlier posts, centralized IT architectures are innately more secure than distributed architectures because they have fewer points of vulnerability. However, this assumes that datacenters are adhering to security best practices. In addition to best practice IT security, datacenters should follow physical security practices such as video surveillance and biometric identification.
The enablement of cloud computing is dependent on these datacenter-related considerations. Perhaps, the most important issue is to ensure that datacenters have the capacity to guarantee the optimum performance of applications that are delivered from remote datacenters, securely.