How CI/CD is Sidetracking Optimization, and What You Can Do About It
High-velocity code changes are making it impossible to optimize infrastructure. But not all is lost in the battle for improved performance.Read more
Asaf EzraAug 12, 2020
How to reduce carbon footprint is the challenge of many in Silicon Valley and around the world. Despite large sustainability programs, businesses are still struggling to become greener and to consume less energy. Technology is a big part of the energy problem, on par with the aviation industry’s fuel emissions as the ICT (information and communication technology) ecosystem accounts for 2% of the carbon footprint. By addressing energy efficiency in power-hungry global data center farms, technology can also be part of the solution. Data centers already account for over 1% of all global carbon emissions, projected to increase as demand for compute rises.
Consuming power 24/7 for data processing servers, network communications, data storage, IT infrastructure, and cooling systems, data centers are estimated to be 1.1 to 1.5% of the global electricity use today to over 273 terawatt-hours (TWh). As data center workloads have grown 6-fold since 2010 for artificial intelligence, connected energy, distributed manufacturing, and other heavy-processing technologies, 18 million global data center servers fill the demand according to 2016 estimates. With the fast pace of change, energy efficiency improvement can help minimize energy use and ensure high power use effectiveness (PUE).
In the continuous data center flux and shifting staff priorities, there’s inherently a lot of inefficiencies in data center servers – both related to server usage and the number of idle servers. Server utilization rates are estimated to be between 10-15% for internal corporate data centers, 20-25% for cloud service providers and 45-50% for hyperscale data centers. Most of the lost PUE is related to idle servers. This creates the opportunity for large improvements in energy efficiency of data center servers. To date, data centers manage their carbon emissions by implementing performance improvements in hardware and silicon and improving cooling infrastructure. With rapidly increasing server demand, this may no longer be enough to keep carbon footprint levels in check. The time is ripe for new optimization technologies to fill the power gap and to drive change for energy efficiency.
Optimizing servers for individual application demands and challenges can help data centers to save energy, power, and electricity. Data centers aren’t using the full potential of virtualization, cloud technology, or related resources; they’re overprovisioning, allocating applications inefficiently, and managing infrastructure inefficiently. By introducing new software performance optimization, data centers can cost-effectively impact carbon footprint.
These new software-based optimization approaches vary in how they balance cost/performance tradeoffs and how easily they can impact data center energy efficiency. Resource management and workload orchestration methods automatically match workloads to resources for provisioning, placement, and sizing. Configuration tuning optimization methods provide automated parameter modification across the technical stack for determining the best workload configuration.
Application-driven resource management optimization is the newest real-time approach that adapts to specific workload needs in the moment, continuously optimizing performance in the OS and kernel. These optimization methods reduce power consumption in servers by boosting server utilization rates to over 80% and removing idle servers, without sacrificing availability. The result is less server waste, less electricity usage, and less physical rack space and more total energy savings with better PUE.
Many existing programs and best practices in data centers focus on cost reduction such as cloud hygiene, rightsizing, automation, and discount purchasing. By leveraging new optimization approaches in software, businesses can more easily and cost-effectively reduce their carbon footprint with significant energy efficiency impact. It’s worthwhile to investigate integrating new optimization upgrades as another step in embracing sustainability.