Welcome to the first installment of our three-part blog series on application-level sustainability. In this series, I will explore challenges and opportunities associated with IT operations sustainability in a hybrid cloud environment, and I will explain how to proactively optimize the carbon footprint of your applications.
Karim Benhabeje (Cisco Sales Enterprise Strategist) authored this blog in collaboration with Cisco colleagues Greg Page (Cisco Sales Enterprise Strategist); Chris Adams (Cisco Customer Experience (CX) Principal Architect); Ewald Munz (Head of Manufacturing, Automotive and Sustainability EMEA at Splunk); Cody Kai Hartsook (Cisco Labs Software Engineer), and Mathilde Durvy (Cisco Innovation & Program Leader). Together, we wanted to delve into the context and constraints faced by current IT operating models. In subsequent parts of this blog series, we will introduce the ‘Workload Carbon Optimiser’ (WCO) application concept and discuss its technical design and implementation.
A bit of context: the growing energy demand of data centers
In our rapidly evolving digital age, it is estimated that data centers (DCs) worldwide account for approximately 1-1.5% of total global electricity use. In 2020, the DCs and data transmission networks that underpin digitalization accounted for around 330 metric tonne CO2 equivalent (CO2e) (including embodied greenhouse gas (GHG) emissions), equivalent to 0.9% of energy-related GHG emissions (or 0.6% of total GHG emissions) according to the International Energy Agency (IEA).
To further complicate the challenge of GHG emissions optimization and reduction, according to Gartner in 2023, 81% of enterprises already use multiple cloud providers to meet their evolving IT needs.
Leveraging expert guidance and automation to implement energy and carbon-saving measures can help optimize workload placement and maximize GHG emissions reduction. Adopting this approach can support business outcomes such as cost reduction, improved consumer perception and corporate brand reputation.
The shift towards sustainability in computing
Application-level sustainability focuses on advancing sustainability in applications and corresponding workloads where they reside (such as Public Cloud, Edge DC, Corporate DC). Business and application owners need visibility into their energy usage, along with related GHG emissions of both their applications and cloud native and edge native workloads, to advance sustainability in operational practices and cost savings goals.
Recent efforts to advance sustainability have focused on powering DCs via renewable energy sources. This includes installing solar panels, wind farms, and other renewable energy projects – locally to the facility – to generate a portion of the electricity needed for their IT and OT operations. Another common strategy is to engage in renewable power purchase agreements (PPAs) with electricity providers. For example, Amazon, Microsoft, Meta and Google are the four largest purchasers of corporate renewable PPAs, having contracted almost 50 GW, equal to the generation capacity of Sweden according to the IEA.
However, two major evolutions have impacted the potential of these strategies to advance private r DCs sustainability. On the one hand, adoption of cloud technologies (especially public clouds) means that enterprises have relatively little influence on hardware, planning or energy sources used by the public cloud service provider. On the other hand, sustainability is, today, a key business objective on the C-suite agenda and is predicted to drive cloud purchase decisions by 2025 according to Gartner. Business departments have environmental, social, and governance (ESG) goals, for digital services, as part of wider corporate ones to reach ‘net zero’ GHG emissions.
KPIs for tracking sustainability in data centers
Tracking appropriate Information and Communications Technology (ICT) infrastructure KPIs is crucial when monitoring and evaluating DCs improvements in line with corporate net-zero goals.
Typically for customer owned DCs the following metrics are measured:
- Power Usage Effectiveness (PUE): PUE is the ratio of total energy consumed by a DC to the energy consumed by IT equipment. It helps assess the energy efficiency of a DC and identify areas for improvement.
- Cooling Efficiency: Tracking metrics such as cooling efficiency (e.g., cooling load, cooling capacity, and temperature differentials) helps optimize cooling systems, reduce energy usage, and improve overall efficiency.
In addition, ICT infrastructure teams running their own DCs today typically use the following resource-level metrics:
- Resource Utilization: Tracking the utilization of computing resources, such as CPU, memory, and storage, helps understand and then optimize resource allocation, reduce waste, and improve overall efficiency.
- Server Utilization: Monitoring server utilization rates helps identify underutilized servers that can be consolidated or decommissioned, leading to energy savings and improved energy efficiency.
- Server Refresh Cycle: Tracking the average age of servers and the frequency of server refreshes helps ensure the use of energy-efficient, modern hardware and enables organizations to repurpose or recycle older, less efficient equipment.
- Virtualization Ratio: Monitoring the ratio of virtualized servers to physical servers helps assess the effectiveness of server virtualization initiatives, which can lead to improved resource utilization and energy efficiency.
- CO2e: Measuring and monitoring workload GHG emissions, and converting them to CO2 equivalent, helps organizations track progress in reducing GHG emissions.
- Energy Usage: Software management capabilities that enable visibility, insights, and action on energy usage.
- Renewable Energy Usage: Tracking the percentage of renewable energy used in DC operations provides insights into the organization’s sustainability and GHG emissions reduction goals.
Traditional ICT infrastructure metrics can help provide a holistic view of the DC’s sustainability performance, identify areas for improvement, and help organizations make data-driven decisions to enhance energy efficiency, reduce environmental impact, and make progress toward sustainability goals.
Challenges of reporting standardization across cloud providers
Even though KPIs might have been properly established and measured, enterprises can face several challenges for sustainability reporting in multi-cloud deployments. These include:
- Standardization and Data: Each respective cloud service provider may use different reporting metrics, making it challenging to consolidate the information. There is no single methodology or model for data collection across multi-cloud providers that would provide a common reporting format with standardized metrics.
- Transparency and Traceability: Public cloud providers located in different parts of the world draw electricity from grids that provide a constantly varying mix of renewable energy sources and carbon-based fuels. It can prove challenging to reliably tell what this mix is at a given point in time.
- Complexity: Hybrid cloud, multi-vendor, multi-domain environments add to the complexity for businesses that are trying to report on GHG emissions. IT Operations, sustainability officers and application owners must deal with this complexity, including dealing with a plethora of different reports, or access to data issues. Advancing visibility into application and workload sustainability requires a new approach.
How observability can help improve your visibility
Building a system that can report on sustainability metrics is an important baseline for many sustainability initiatives. To support this, enterprises can benefit from an observability solution across the applications landscape and entire stack.
Observability uses data points like metrics, events, logs, and traces (MELT) to provide insights into system performance and health through a platform such as Splunk Observability Cloud (SOC). This can help improve operational efficiency, including optimizing and reducing the organization’s application landscape carbon footprint.
Looking ahead
In part 2 of this three-part blog series, we will introduce the application concept: ‘Workload Carbon Optimizer’ (WCO). This will be followed by part 3 that will go into greater technical detail on the functional design architecture of WCO validated by Cisco’s own internal proof of concept.
We’d love to hear your thoughts on the challenges and opportunities of hybrid cloud sustainability. What strategies have you found effective in reducing your applications and workloads carbon footprint? Do you have questions or insights about the topics discussed in this blog? Please share your comments below and join the conversation. Your feedback is invaluable as we continue to explore and innovate in the realm of sustainability in IT operations.
