Today’s data center managers must not only satisfy customer demands for around-the-clock availability from anywhere in the world; they must also contend with demands from within their own organizations to help reduce operational costs.
Customers and internal stakeholders alike expect and ask for the same availability as the traditional “plain old telephone service” (POTS). In days gone by, providers such as AT&T engineered their dial-tone service to be available 99.999 percent of the time. This dial-tone reliability has become such a well-known benchmark that it is commonly known as the “five nines” standard, which is the equivalent of having a dial tone available for all but five minutes a year.
However, high expectations translate into increasing pressure on data center managers, who can quickly find themselves on the horns of a dilemma. On the one hand, they must keep their facilities operating at peak performance. On the other hand, they have budgetary concerns from within their own companies because external regulatory bodies are demanding that energy usage be reduced.
This paper is intended for executives who determine their organization’s business strategies and IT policies. If you are looking for ways to decrease data center costs, and are considering a variety of options, you require knowledge of the possibilities that are available as well as the successes that others have had.
A primary question you must answer is “how can organizations continue to provide availability, scale to the needs of their customers, and remain cost-effective?” This single question requires that you look at multiple ways to make your data center more efficient, as well as consider other options, such as offloading some of your applications to the cloud.
The data center is a complex environment. You will need to examine how well you take advantage of virtualization, multi-tenancy, location independence, and optimization. Two other issues to consider are how innovative your IT department is and if you have a good business continuity plan in place. Highly successful data centers spend a great deal of their time and budget on new projects and reengineering their services. And, as Hurricane Sandy brought home to many, a business continuity plan gives you a strategy for dealing with disasters.
Another option is to examine the efficiencies that are potentially available by using the cloud. Disaster recovery is one application that may be well suited to the cloud because it normally uses a small number of resources and only requires maximum resources when a major problem occurs. Of course, you must ask potential vendors many questions, such as what their bandwidth capacity is, what their own disaster recovery plan is, and how they handle information security.
Finally, what technologies and approaches can make your data center more efficient – not just in the short term, but for its entire lifespan? For example, people are discussing how to enforce accountability, which means how to make sure departments are held responsible for the energy they use, new energy-efficient processors, the merits of DC power, and better ways to measure that elusive quality: efficiency.
After reading this paper, you will better understand the current challenges surrounding data centers and some of the concrete benefits that can come from increasing efficiency in your own data centers, as well as how the cloud can also help you to reduce your costs and your carbon footprint. Some of the issues we address are:
- Trends driving data center efficiency go beyond energy. Other factors include how well data centers can accommodate growth, how well critical systems are protected, and how well they can satisfy customer demands.
- Multi-tenancy can include shared IaaS services such as metered usage and identity management as well as the PaaS layer, which may offer such things as application servers and development environments, and might ultimately extend to the SaaS layer.
- Hurricane Sandy in 2012 demonstrated how good continuity planning involves redundant data centers, how the data center is organized in terms of where equipment is located, and how to handle generator systems.
- While true in many cases, claims that cloud providers have enormous economies of scale and therefore will be less expensive than existing IT resources are overly simplistic and do not take into account the kind of company and efficiencies already achieved for the relative unit cost of some resources, such as computer storage.
- Any sort of disaster recovery (DR) plan needs to determine the recovery point objective (RPO) and the recovery time objective (RTO) in order to understand what happens if a particular process or application goes offline. While disaster recovery plans often focus on bridging the gap where data, software, or hardware have been damaged or lost, they should also take into account what happens if personnel is not available.
- While power usage effectiveness (PUE) is a starting point for measuring a data center’s efficiency, other considerations such as carbon emissions, water usage effectiveness, and energy sources are also important when determining how green a data center really is.
- Trends driving data center efficiency
- Increasing data center efficiency
- Improving efficiency with the cloud
- Disaster recovery
- Related issues
- Key takeaways
- List of Resources
- About Dave Ohara
- About GigaOM Research