We often hear specific terms such as 99.99 percent reliability and 300W per square foot. These terms are often associated with data center. In the event of catastrophic failures, communications systems, applications and data center servers can be at risk. Eventually data centers could continue to get denser as new racks and equipments are added to available spaces. It means that we will get extra challenges and we should be able to overcome them with proper methods. There are space and power requirements when we want to establish a scalable data center. Once of the biggest concern is the cooling solution.
Cooling system in data centers should provide redundancy to improve scalability. Without proper cooling, the internal temperature could eventually go higher than the allowed limit. Many data centers have raised floors that allow cold air to be blown through special pipes and directly to the equipments. This is more efficient making the whole room cooler. Cold air can be directed to hottest spots in data centers. We should also consider the proper space requirements. Space can be an expensive asset, especially if data centers are located in urban areas. It is important to define proper layout to maximize the use of space.
Electrical systems should also be able to handle the increasing demands of power. Proper equipments should be used to distribute power from the power-supplying company to the smallest equipment in the data center. Dedicated UPS or uninterruptible power supply should provide temporary power during blackouts. UPS are essential, because brownouts are more likely to happen and computer could shut down, even if the electricity is not completely cut off by blackouts. The amount of wattage needed in data center should be continuously calculated and more UPS modules could be needed in data center to make sure that nothing is disturbed.
When managing a data center, we should consider risks of failure. In this case, failures are often associated with excessive temperatures. We should have good heat removal capability. However, data center equipments have expected lifespan and it is important to replace equipment when it has reached 95 percent of its expected lifespan. If equipments fail suddenly, it could become very difficult to restore the situation. When equipments or servers die, they could bring some critical information with them, even if we have implemented rigorous data backup policies. This is especially true when we are processing real-time data.
Technically, it is possible to address common data center challenges, even in a dense environment. We should make a projection of what will happen when the amount of power per square foot in our data center is getting higher and higher. Good redundancy in terms of cooling and power system is needed. Innovating solutions, such as raised floor could solve many problems and add some flexibility in our designs. We should be prepared for potential failure in our data center and arrange procedures to contain and counter them. Failure on one equipment should be isolated, so they won’t cause catastrophic effects on the rest of the system.