Is there really nothing we can do about soaring data center energy consumption expenditure?
Respond to rapidly rising energy consumption costs through the agility and high controllability of IT architecture
Today, all aspects of data center management are developing rapidly, but the operating budget is hard to keep pace. According to statistics, the total amount of data will double every 18 months, and the load on the application side will double every 2 years. However, it will take 8 years to double the operating budget.
People have been asking the IT industry to do more at a lower cost. In recent years, data centers have increasingly high requirements for dynamic characteristics. Smart devices, big data, virtualization and cloud technology continue to change the mode of data center services. Therefore, flexibility, flexibility and scalability are becoming increasingly important.
Due to these rapid changes, all aspects of data center management have become increasingly complex. Thousands of devices on the internal network are constantly connected and disconnected. At the same time, the mobile computing resource pool is automatically allocated and managed. So, in this super dynamic environment, does it mean that the IT management team cannot effectively identify idle and overloaded resources?
If so, data center energy consumption will continue to soar. According to the latest report of the Natural Resources Conservation Commission, at present, the energy consumption of data centers has accounted for an astonishing 10% of the total global energy consumption
Fortunately, even if the challenges facing the data center are so severe, the IT industry is not powerless.
The discovery of secret weapons
Ironically, today's data centers are mostly composed of resources defined based on software, but the secret weapon to control energy consumption costs depends on hardware instead. Rack and blade servers, switches, power distribution devices and many other data center equipment will provide a lot of energy consumption and temperature information during operation. In view of the huge scale and wide range of hardware equipment in the data center, it is very difficult to collect and apply information resources manually, which also promotes the vigorous development of energy consumption management solution providers.
As a result, data center managers now have many choices. They can use the management console with built-in energy management to let the integrator integrate the energy management middleware software into the existing management console, or independently deploy the energy management middleware solution to obtain the necessary functions.
No matter which method is used for deployment, a complete set of comprehensive energy consumption management solutions can enable the IT management team to view, record and analyze the energy consumption and temperature operation status of the entire data center. The energy consumption and temperature information automatically collected and concentrated can be further drawn into the image information of each space in the data center. These data can be used to analyze and identify the corresponding operation rules, and understand the workload and other changes in the data center.
Clear, visible, recorded and archived power information can enable data center managers to have a basic understanding of power usage, so as to make more favorable decisions on data center planning and optimization.
The most excellent energy consumption management solution improves the optimization measures to a higher level by combining the automatic monitoring and recording system with the real-time control function. For example, if conditions permit, or in an appropriate period of time, limit values can be set for some servers or racks to limit power within a certain range. For servers that have been idle for more than a specific length of time, they can be automatically converted to a more power-saving sleep mode. The use of power can be flexibly distributed according to the business focus. This function can also effectively extend the maintenance time of standby power supply in case of power failure. An excellent energy consumption solution can even dynamically adjust the dominant frequency of the server to minimize energy consumption without negatively affecting the quality of service and application performance.
Data centers that care about energy consumption level will make full use of these functions to achieve a series of operational goals, such as accurately planning workload, saving operating costs, extending the life of data center equipment, and achieving relevant environmental protection specifications.
Application and effect proof
In practice, the customer revealed several common deployment motivations and provided in-depth understanding of the effect and degree of the overall energy consumption solution and its best practices.
Energy consumption monitoring: Identifying and understanding the peak period of energy consumption is the source of many companies introducing energy consumption management solutions. By understanding this information, customers can reduce their electricity consumption by 15% during peak hours. Even when the demand for electricity increases sharply during peak hours, they can still reduce the monthly electricity bill of the data center. Energy consumption monitoring is also used to accurately charge users of hosting services and other services.
Increase cabinet density: the area of the computer room is also one of the factors restricting the expansion of many data centers. In the absence of real-time information, the traditional fixed power supply mode relies on the rated value of the power supply or the non rated value based on experimental data. Real time power monitoring has generally proved that the actual power consumption is usually far below the rated value. After the monitoring and power capping mechanism is added, the data center can more actively increase the number of racks, and increase the density of server cabinets by 60% or even more than 80% under the same power range.
Find idle or low utilization servers in time: During peak workload, similar "ghost" servers occupy up to half of the power consumption. The energy consumption management solution shows that at any point in time, such servers account for 10% to 15% of all servers in the data center. This information can help data center managers better consolidate and virtualize to avoid this waste of power and space.
Identify potential faults in advance: power monitoring and control can not only monitor dangerous hot spots and automatically give an alarm, but also increase the operating time of standby power supply by 15% and extend the business duration by 25% in case of power failure.
Advanced temperature control: the real-time temperature information acquisition function can help draw the intuitive heat image of the data center without adding expensive temperature sensors. The use of temperature images can significantly improve the monitoring effect of the data center, and the monitoring range can effectively cover the entire floor, even each device These temperature maps can also improve capacity planning and help avoid under or over cooling. Through stronger visual images and setting limit values, the data center management can more confidently improve the operating environment temperature of the computer room. 5% to 10% of the cooling cost can be saved when the ambient temperature is increased by 1 ℃.
Effectively balance energy consumption and performance: The data center can effectively reduce power consumption by 15% to 20% without affecting or almost without affecting performance by changing to a more intelligent processor.
It's time to take energy consumption seriously
The hardware facilities of the data center are still important. The software responsible for allocating resources to applications and services continues to evolve, which requires real-time fine monitoring of hardware devices. The introduction of energy consumption management solutions just makes this monitoring mechanism and energy consumption and temperature control possible, and allows IT management departments to effectively control power resources. Power expenditure has already become the largest single expenditure on the operating budget table.
Software and middleware solutions can enable data center managers to effectively monitor the hardware and its operation status, thus making the automation process faster, safer and more cost-effective, and avoiding a sharp rise in power costs. In today's energy deficient data centers, energy aware virtual machine migration and work scheduling should become an industry standard.
About the author:
Jeff. Mr. Klaus is the general manager of data center manager (DCM) solutions of Intel Corporation. At Intel, Jeff. Klaus has more than 13 years of team management experience. His team is developing energy consumption and heat management middleware products. Jeff. Mr. Klaus holds an MBA from Boston University.
