Choosing the Right Data Center for Disaster Recovery Planning
How to select the best data center for your business using nine key factors
Mark Noonan, Vice President of Sales, Data Foundry (Winter 2011)
Choosing a facility designed to securely house your company's computer systems, networking, and other technology components is a crucial decision for any business. Companies of all sizes depend on the stability and reliability of IT infrastructure to operate successfully. Even one moment of network disruption can wreak havoc on a company's revenue and customer relationships. Businesses need a robust and comprehensive disaster recovery plan to ensure that operations run smoothly in case of disaster. Data centers are an important component to any disaster recovery plan, serving to house core network and infrastructure operations and providing space for the people that run the organization.
Companies need a disaster recovery plan to limit the crippling effects of a disaster. Disasters can include human error or natural occurrences such as hurricanes, flooding, or earthquakes. Many large companies spend 2-4 percent of their dedicated IT budgets on disaster recovery planning. This investment helps avoid additional losses to the business associated with IT infrastructure failure. It is important to carefully assess the data center components involved in disaster recovery planning to ensure that time and money is well spent. Here are nine key elements to consider when choosing the right data center.
A data center that offers an optimal environment to keep the facility running with minimal down time and low operating costs is a high priority. An optimal environment has a low risk of natural disasters, a favorable business climate, and a rich technology talent pool. Companies are shying away from coastal regions because of hurricanes, earthquakes, and wildfires, and from high-profile metro areas because of the threat of terrorism. The best cities for data centers are in the central and southwestern United States due to the lower incidences of natural disasters, according to a 2010 study by the Boyd Company.
Carrier-neutral network access is also important. Austin, Dallas, Atlanta, and Raleigh-Durham are the some of the most popular data center cities, as they are rich with fiber options. Other criteria to consider when selecting a data center location are the price and reliability of power in the area. Incentives, such as cheaper taxes and rates from local governments and utilities, also influence the data center site decision.
A reliable infrastructure eliminates points of failure and employs total redundancy where necessary. The mechanical plant within a data center includes power, electrical, and cooling systems, which keep it operating smoothly 24/7/365. The mechanical and electrical systems should be designed with multiple levels of redundancy. For instance, data centers often have multiple generators to ensure backup power is available in case of a utility outage.
Building a data center that is fully redundant at every level can be cost-prohibitive, and some data center operators choose not to make this investment. Review the data center system design and the provider's track record on outages, availability, and service level history. The data center provider should describe in detail ongoing maintenance for any of its critical systems and standard procedures for correcting an issue.
3. Redundant Utilities
Data centers often have redundancy built into the primary power, water, and network utilities. Data centers should provide independent, reliable, scalable, and resilient power with diverse feeds entering the facility. Diverse feeds act as added insurance to keep operations running seamlessly if one feed is compromised.
Ideally, choose a data center that not only has diverse power feeds, but with feeds that originate from more than one substation. A data center should also have two separate water utility feeds. At the network level, data centers should be designed with dual entrances for all providers and carriers, which further increase reliability and uptime.