Data Centers are the backbone of today’s IT world. Growing business, demand that the data centers operate at maximum efficiency. However, building data Centers, maintaining and running them involves a lot of operational expenses for the company. It is important for companies to look for options that can help them lower Opex for their data centers. Proper capacity Planning, advanced monitoring techniques, and predictive analysis can help companies to achieve these goals and help improve business growth. Real-time monitoring helps data center operators to improve agility and efficiency of their data centers and achieve high performance at a lower cost.
Today’s digital world requires constant connectivity, which in turn requires all time availability. But there could be several things that could cause outages – like overloaded circuit chip, air conditioner unit malfunction, overheating of unmonitored servers, failure of UPS (uninterrupted power supply) and power surge. So how do we ensure availability? Implementing DCIM (Data Center Infrastructure Management) technologies can help you improve reliability. DCIM systems monitor power and environmental conditions within the data center. It helps in building and maintaining databases, facilitate capacity planning and assist with change management. Real-time monitoring helps improve availability and lower Opex.
Servers and electronic devices installed in Data Centers generate a lot of heat. Overheated devices are more likely to fail. Hence, data centers are usually kept at temperatures similar to refrigerators. Thus most of the power in a data center is consumed for cooling purpose. There are various techniques and technologies that data center operators can implement to save energy. Recent strategies like free cooling and chiller-free data centers, expand the allowable temperature and humidity ranges for data center device operations. Implementing these strategies help save energy costs. A telecommunication giant CenturyLink had an electricity bill of over $80 million in 2011 which made them think of a solution to lower this cost. CenturyLink implemented a monitoring program. With this monitoring program, their engineers were able to safely raise the supply air temperatures without compromising availability and with this solution CenturyLink was able to save $2.9 million annually.
As per ASHRAE (American Society of Heating, Refrigeration and Air Conditioning Engineers) new guidelines, the strategies like free cooling and chiller-free data centers can offer substantial savings and one might expect data center operators would make use of these seemingly straight forward adjustments. However, as per a survey, many data center operators are not yet following these techniques and average server supply air temperature for the data center is far cooler than ASHRAE recommendations.
Most of the data centers are provisioned for peak loads that may occur only a few times in a year. Server utilization in most of the data centers is only 12-18% or may peak at 20%. However, these servers are plugged in 24x7x365. In summary, though the servers are idle they are drawing the same amount of power that other operational servers are drawing. Power distribution and backup equipment implemented in data centers also cause substantial energy waste. Similar to cooling strategies, most of the owners employ alternate strategies to improve power efficiency. However, most of them are on the computer side. Increasing density of the IT load per rack, with the help of server consolidation and virtualization, can offer substantial savings, not only in equipment but also in electricity and space. This is an important consideration when a data center is located in constrained energy supply or electricity situation in the context of high real estate prices, as in most of the urban areas.
Increasing density leads to concentrated thermal output and needs modified power requirements. The effective way to maintain continuous availability in high-density deployments is real-time monitoring and granular control of the physical infrastructure. Power proportional computing or matching power supply to compute demand is the recent innovation that few of the operators are using to improve energy efficiency. Few operators use dynamic provisioning technologies or power capping features already installed on their servers. However, raising inlet air temperatures causes the risk of equipment failure. Without an in-depth understanding of the relationship between compute demand and power dynamics, implementing power capping increases the risk of the required processing capacity not being available when required. Without real-time monitoring and management, there is a high risk of equipment failure in a data center.
Real-time monitoring helps businesses get critical information to manage possible risks in the data center. Monitoring helps improve efficiency and decrease costs, enabling businesses to have availability and saving. They can lower Opex and still maintain high availability.
With the help of Real-time monitoring, a small issue can be spotted, before it becomes a large problem. In a smart data center, several thousands of sensors across the facility collect the information regarding air pressure, humidity, temperature, power usage, utilization, fan speed and much more – all in real time. All this information is then aggregated, normalized and reported in a specified format to operators. This allows operators to understand and adjust controls in response to the conditions – to avoid failures and maintain availability.
Monitoring has lot many benefits. Monitoring data can be used by cloud and hosting providers to document their compliance with the service level agreements. Monitoring data allows operators to automate and optimize control of physical infrastructure. Real-time monitoring gives visibility at a macro and micro level, for businesses to improve client confidence, increase data center availability, energy efficiency, productivity and at the same time reduce their operational expenditures by optimizing data centers with the help of monitoring data.