Reducing Data Center Power Demand with Cisco UCS
Look at the operating costs for your data centers and you’ll likely see a big amount for the electrical power to run the servers, storage, networking components, and cooling systems. Since power consumption is an area where even small changes can add up to big savings over time, we want to take advantage of every power-saving feature we can find. And we’ve found many of those features in the Cisco Unified Computing System (UCS) servers, which we now deploy as the standard in our data centers worldwide.
Of course, the primary way Cisco UCS reduces our electricity usage is by hosting multiple virtual servers on a single physical server, which reduces the amount of installed but unused server capacity. The Cisco UCS design also reduces direct power consumption and demand for cooling (which consumes additional power) by incorporating industry-standard, low-power processors and memory modules.
Several Cisco UCS features are focused specifically on power and cooling efficiency. For example, direct delivery of 12V power to high-efficiency, point-of-load converters on the Cisco UCS server motherboards reduces the power losses that would be associated with multiple voltage conversions.
The power supply units (PSUs) in a Cisco UCS chassis are rated to the 80Plus Platinum standard, providing over 94 percent peak efficiency and a broad efficiency range. The power supplies can be configured in a nonredundant, N+1 redundant, or grid-redundant design to deliver continuous power. In addition, Cisco UCS Servers are certified to the EPA Energy Star 2.0 standard, where applicable, demonstrating the energy efficiency of the Cisco UCS servers.
Cabling also plays a role. When compared to 10GBASE-T technology, the SFP+ CU cables in Cisco UCS servers use less power. And by using Fibre Channel over Ethernet (FCoE) technology, Cisco IT is able to reduce server cabling and the associated power consumption.
When it comes to reducing the power demand related to cooling, a motto for Cisco UCS servers might be “better cooling through simplified design.” The server chassis allows straight-through airflow without the bends and turns that reduce the cooling impact. With a 63-percent open midplane, the Cisco UCS chassis also allows us to direct airflow to the components that most need it for optimum cooling within the server and chassis.
Complementing these cooling features are Cisco UCS capabilities for active thermal monitoring, such as the variable fan speed feature that automatically reacts to changes in the thermal output of the chassis. Thermal sensors are used on all major server components and the sensor data is aggregated by the Cisco Integrated Management Controller and the Cisco UCS Manager software for the entire system. Additionally, automated alarms alert our operations team when things get too hot.
Features for directly managing power consumption are also helpful to us. The Cisco UCS servers have power measurement sensors at both the blade and chassis level, which allow us to measure the actual power consumption for each server. And with capabilities for power capping, we have the option to dynamically limit the overall power consumption of a Cisco UCS system and send more power to the blades that need it.
By taking advantage of all these Cisco UCS features, we know we’re reducing power usage in our data centers. Learn more about how Cisco UCS may reduce power consumption in your environment by exploring these resources:
- Cisco UCS Power Calculator
- Cisco UCS Power Efficiency Overview Report and comparison reports for Dell, HP, and IBM servers
What strategies do you use to save power in your data center?