Pages Menu
TwitterRssFacebook
Categories Menu

Posted in Articles

Data Centers; Powering the Web

Data Centers; Powering the Web

For more interesting articles on data centers, please visit:  http://www.processor.com

 

The amount of power data centers use to keep dedicated servers running and cooling the building is sky high and is a big part of a data center’s daily cost. According to the US Department of Energy, as little as 15% of the originating energy sources is actually used for the equipment used within a data center. Most however, use around half of all energy on cooling equipment alone while the rest is used on the dedicated server hosting operations and tasks.

Data center energy requirements have more than doubled since 2001 but what many web hosting providers are doing in order to lower costs and at the same time help lessen the strain on the environment is by simply removing unused servers, virtualization which can help significantly on idle time, or adding newer more energy efficient servers and replacing older dedicated servers, improving energy efficiency via infrastructure with better airflow management or via better fans and chillers or even free cooling such as building data centers in cooler climates. A good example of this is the new data center located in Prineville, Oregon where Facebook and Apple have their new data centers located providing a very low Power Usage Effectiveness (PUE) because of the cooler and less volatile temperature variations. (Taxes were also a big incentive for the two companies to move to Prineville as Oregon has no sales tax). Having the option to use cool outside air from Mother Nature to keep an entire building cool from overheating dedicated servers can provide a huge amount in savings. Why not located data centers to places such as Greenland for example? Where are the fiberoptic connections? Fast and redundant connections are vital in the daily operations of any data center. Hiring qualified staff to live and work in such climates would indeed also be a challenge!

A lot of east coast data centers can encounter trouble via storms in that part of the country, however data centers on the east coast are still in high demand because they are not on any fault lines, aka earthquakes so are presumed a safe bet for uptime and reliability and are well connected to the Internet via high speed fiberoptic connections. Another popular location is Texas which does not encounter natural disasters but cooling costs can run high as the temperature in that part of the country is very hot. However, land and operating costs (taxes) in southern and central parts of the U.S.A. is at times considerably cheaper hence the reason for still popular demand for data centers in that area. Cheap power is very high on the list on deciding where to build any new data center within the United States.

Creating a balance between natural disasters, costs and fast redundant connections are the main deciding factors to building and running or co-locating a dedicated server in a data center anywhere in the world.

In some cases, businesses do not have much of a choice in deciding a location to use as latency may simple be the deciding factor. Companies that may focus on this aspect greatly are traders on exchanges or game servers. Sometimes it can simply be that a company wants to be nearby a data center they are co-locating as the “hands and eyes” option is important to them where they can just pay a visit.

What to look for in a data center

Every data center must have backup power, such as diesel fueled generators or UPS’s to ensure power is provided to the entire building housing the dedicated servers 24 hours a day 7 day a week. Better yet, finding out how long a data center can continue running on this backups power can be important to know.

Along with power security a data center must be able to provide bandwidth and traffic that can fluctuate on an unpredictable basis and be able to handle spikes in traffic requirements.

A data center must also have a strong Service Level Agreement that states in an event of hardware or network failure what takes places to resolve the downtime and if/when hardware is replaced. A good indicator of dedicated server network uptime is an APS350 or higher certification.

24/7/365 support and security plays a big part in hosting with a data center. If you are accessing your dedicated servers remotely you want to be sure security is always in order and any support requiring hands on access is available at any time of the day. If you cannot access a server remotely this is vital. An alternative or backups solution to this could be to see if the data center offers KVM units or at least remote reboot ports. Some dedicated servers such as the Intel Xeon E3-1230 3.3Ghz QC have Supermicro IPMI come with these features already.

Last but not least, any data center must be financially stable and secure including the history of the data center and how long it has been operating. If it goes under, you don’t want to be in the middle of the turmoil and trying to get your servers or data out of there as quickly as possible. Promises and verbiage will not make up for any current lawsuits or financial trouble. If the data center is new is it backed by strong finances or are there other data centers built by the same company with good records.

 

For more interesting articles on data centers, please visit: http://www.processor.com