Data Center Construction & Importance Of Reliability

Explore Requirements For Modern Data Center Construction

Data center construction began to be an important consideration for many businesses, schools, and other consumers of computing power since the 1970s. At that time, large mainframes would take up an entire room. Another room could be taken up with the data backup tape drives. These early data centers had two things in common: they used a lot of electricity and they generated a lot of heat.

Interestingly, the computer power that used to be contained in a single room is less than is found on the common garden-variety laptop today. Similarly, the data in all the tape drives in the second room could be held on a single modern flash backup drive. However, the need for computing power has grown at least as quickly as the industry’s ability to shrink the size of computer hardware. Now that computing has become a critical requirement for many companies, data centers are built as a part of many campuses. Similarly, with the increased use of distributed computing such as used in cloud computing, stand-alone data centers have been constructed to supply the need.

Data center construction has some specific hardware and software requirements to be reliable and efficient. The main hardware in a data center is a server. Groups of servers are mounted in rows forming corridors, like the arrangement of many libraries. The standard 19-inch rack-mount servers, also known as 1U servers, are one implementation. Servers can also be constructed as a silo that can occupy several square foot of floor space. When servers fail, the entire group can be replaced at a time.

The key to an effective data center is reliable communications. This means that both the servers and their power supplies are designed in a redundant fashion. Two storage servers may maintain the exact same data, in order for the backup to step in if the primary unit has any troubles. Computing servers may operate in parallel in order to make sure that the computing stream is uninterrupted in the face of any hardware malfunctions. The other component of reliability is environmental control. This includes air conditioning, fire control, and equipment security devices. In addition, the main power routing must be carefully controlled. For some large data centers, the main power would be enough to run a small town. In the same way, data centers must be monitored for air pollution. They expel gases in the form of diesel exhaust under normal operation.

The software and control of the data center is also important. The regulatory body of the Telecommunications industry Association (TIA) generates specifications for data center construction. The standard is published in TIA942 Telecommunications Infrastructure Standard for Data Centers.

Data center construction follows one of four tiers, in order of increasing reliability. Tier 1 offers little hardware redundancy, and guarantees a maximum down time of 3290 ppm (parts per million). Tier 2 adds redundancy components with a maximum down time of 2590 ppm. Tier 3 guarantees multiple, unique distribution paths. Each component must have redundant power supplies. The downtime is 180 ppm max. Tier 4 requires environmental controls be redundant as well, and has a maximum downtime of 50 ppm.

As businesses specify the requirements for their own data center construction, or the specifications for the cloud they plan to use, they match their requirements to the capacity and reliability of the data centers they are considering.

This entry was posted in Uncategorized. Bookmark the permalink.

Comments are closed.