How Holaluz reduced their AWS cloud-based power control center emissions by 40%

In this post, Sergi Teixidó tells us about how Holaluz achieved a reduction above 40% as measured with the Cloud Carbon Footprint tool by Thoughtworks, with the help of AWS Enterprise Support.

We work closely with our Technical Account Manager to address each new challenge we face through AWS Enterprise Support entitlements. They connect us with key AWS subject matter experts to expedite the adoption process and achieve better results. In the last 6 months they have helped us in our path to reducing the CO2 emissions linked to our AWS usage, which is a key objective for HolaLuz. Achieving a reduction of more than 40% as measured with the Cloud Carbon Footprint tool by Thoughtworks. -Sergi Teixidó – Holaluz Head of Infrastructure and Platform

The mission

Holaluz was founded to address the global climate crisis by accelerating the transition to renewable energy. As part of this mission, the company includes a sustainability perspective when optimizing and improving its applications.

In 2022, after experiencing substantial growth from deploying Spain’s first cloud-based power control center, Holaluz strategically embarked on an Enterprise Support journey to fine-tune its cloud infrastructure. This move aligned with the company’s sustainability goals by making sure of cost-efficient operations and mitigating environmental impact by reducing unnecessary resource usage.

From December 2022 to May 2024, Holaluz, guided by Enterprise Support, led an optimization journey, following the best practices of the Sustainability Pillar of the AWS Well Architected Framework to reduce their AWS footprint. Since then, Holaluz has made sustainability optimization part of their sustainability operations.

The optimization for the sustainability journey on AWS

AWS is committed to running its business with minimal environmental impact. We also strive to enable our customers to use the cloud’s benefits to better monitor and optimize their IT infrastructure. As reported in How moving to AWS cloud reduces carbon emissions, our cloud infrastructure is 3.3 times more energy efficient than the median European enterprise data center.

However, sustainability is a shared responsibility between AWS and our customers. AWS optimizes the cloud’s sustainability (sustainability OF the cloud), while customers are responsible for sustainability in the cloud, meaning they must optimize their workloads and resource usage (sustainability IN the cloud).

Sustainability in the cloud is an ongoing effort focused primarily on energy reduction and efficiency across the components of a workload. This process involves understanding the current state, identifying areas for improvement, selecting targets, testing enhancements, adopting successful improvements, quantifying success, sharing learnings for replication, and repeating the cycle.

This continuous effort needs to be deeply embedded in a company’s culture. Teams should feel ownership and accountability for this process. By making sustainability a key metric or KPI, it becomes more ingrained as a standard practice across teams. It is crucial to involve stakeholders from various departments, including operations, finance, and leadership, to make sure that your sustainability initiatives integrate into an organization’s long-term plans and decision-making processes.

AWS provides best practices for sustainability on the Sustainability Pillar of the AWS Well-Architected Framework. Each workload and business need differs, with varying resource usage and environmental impact. To get started:

1. Understand your workloads’ impact and alignment with business goals. Identify resource-intensive deployments and frequently used resources. Evaluate these hotspots for opportunities to improve resource usage.
2. Evaluate and prioritize: Estimate each opportunity’s impact, implementation costs, and associated risks. Prioritize improvements based on the greatest anticipated impact with the lowest costs and acceptable risks. Incorporate them into resource planning and development roadmaps, and measure improvements by monitoring resource consumption.
3. Deploy and measure: Once improvements are deployed, measure their impacts on your workloads. Proxy Metrics provide near real-time, fine-grained insights into workload efficiency.
4. Automate and repeat: Optimization for sustainability is a continuous process. Establish mechanisms to continuously analyze usage metrics and prioritize optimization.

How Holaluz is optimizing for sustainability

Holaluz’s mission is to build a more sustainable world. It was logical for them to embark on the process of optimizing for sustainability with the goal of reducing their AWS workloads’ footprint, including carbon footprint reduction as a business goal.

Following this mission and their sustainability business goals, Holaluz started the continuous process of optimizing their workloads for sustainability. Holaluz defined this process, understanding that it needed to be iterative and continuous, aligned with their business mission, and include a cultural aspect.

Holaluz’s optimization journey has had two phases: a first phase where they undertook an optimization process, and a second, current phase where they have made optimization part of their daily operations.

First phase: optimization of the workloads

In the initial phase, Holaluz conducted a thorough analysis of their workloads, identifying opportunities for significant emission reductions with minimal effort.

Their approach began with removing or refactoring workload components with low or no use. By leveraging AWS Trusted Advisor, they pinpointed unused Amazon Elastic Compute Cloud (Amazon EC2) instances and idle Amazon Relational Database Service (Amazon RDS) instances, enabling them to deactivate unnecessary resources.

Then, Holaluz addressed non-production workloads active during off-hours by implementing AWS Instance Scheduler. This solution automatically shuts down these workloads outside of working hours while allowing manual activation when needed. Sergi noted that this implementation was relatively direct.

To make sure of the smooth adoption of these changes, Holaluz created a Slack integration for automatic machine activation. This user-friendly approach maintained developer workflow efficiency while aligning with the company’s sustainability goals. As a result, the non-productive workloads now cease operation after hours, with developers fully on board with the initiative.

Next, Holaluz focused on selecting instances with the least impact. Recognizing the energy-saving potential of AWS Graviton, Holaluz migrated their Amazon RDS instances to Graviton. This change was seameless for most of their databases, except for those needing version upgrades to align with Graviton’s minimum requirements.

Although making sure that the teams updated their databases to the latest PostgreSQL version posed a challenge, Holaluz viewed it as an opportunity for overall system modernization. Ultimately, they successfully migrated 80% of their databases to AWS Graviton, realizing a 50% reduction in emissions and achieving a comprehensive database update.

Second phase: continuous sustainable operations

In the second phase, Holaluz focused on automating optimization processes to maintain stable AWS usage and emissions over time.

To follow data management best practices, they implemented AWS Backup, achieving an automated backup strategy. This was complemented by establishing policies to manage the lifecycle their data, to prevent unnecessary data accumulation and reduce the carbon footprint associated with storage. Furthermore, they optimized Amazon S3 data storage through Amazon S3 Intelligent-Tiering.

AWS Enterprise Support played a crucial role in simplifying the implementation of a comprehensive backup strategy, including cross-Region and cross-account replication. As a result, Holaluz now boasts an optimized disaster recovery (DR) strategy that minimizes data duplication and accumulation.

To make sure of efficient hardware usage, Holaluz uses Amazon CloudWatch through the AWS Command Line Interface (AWS CLI) to identify overprovisioned databases and scale up or down their Amazon RDS instances monthly to align capacity with business needs. In 2024, they began using AWS Compute Optimizer to determine optimal resource configurations for Amazon EC2 and Amazon Elastic Block Store (Amazon EBS), with plans to extend its use to Amazon RDS. For Amazon EC2 resources, Holaluz leverages AWS Trusted Advisor alerts that highlight EC2 instances with low CPU usage or network I/O over a prolonged time period.

Throughout this process, Holaluz found these measures easy to implement, allowing them to maintain stable AWS usage and identify anomalies in real-time. Their commitment to sustainability has not only reduced their carbon footprint but also improved their overall system efficiency and resilience.

Next steps

Holaluz uses Amazon EC2 Auto Scaling and AWS Auto Scaling for Amazon Elastic Container Service (Amazon ECS) to automatically scale in and out computing resources according to demand. However, only 20% of AWS Auto Scaling groups involve productive workloads for now.

After participating in an AWS GameDay organized by AWS for Holaluz in June 2024, Holaluz has realized implementing AWS Auto Scaling needs less effort than expected. Implementing dynamic scaling is now part of their short-term roadmap.

Making sustainability part of the team’s culture

Holaluz promotes sustainability by making sure that teams understand the cost impact of their infrastructure deployments. They use Terraform’s infracost to notify the person deploying new infrastructure about the estimated cost through Atlantis executions, keeping everyone aware of the impact of their changes.

To increase transparency, Holaluz has implemented an extensive tagging strategy and has deployed customized cost dashboards in Amazon QuickSight for each team, fed by the AWS Cost and Usage Report. These dashboards provide real-time, granular visibility into AWS spending, broken down by services and resources used. This enables better accountability, informed decision-making, and opportunities to optimize usage and reduce waste.

The results

Holaluz have worked closely with their Technical Account Manager and key AWS subject matter experts to expedite the adoption process and achieve better results. From January 2023 to June 2023, Holaluz achieved a reduction above 40% as measured with the Cloud Carbon Footprint tool by Thoughtworks.

Figure 1: Holaluz optimization’s first phase results, according to the Cloud Carbon Footprint tool by Thoughtworks. The figure shows how switching off non-productive workloads during the weekend reduces by 10% relative emissions.

Thanks to the different mechanisms implemented by Holaluz to monitor AWS usage, they have maintained stability during the last year.

Figure 2: Holaluz’s AWS usage during the last year.

Thanks to the efforts described in this post, Holaluz has gained granular control and visibility over their AWS usage, enabling them to identify optimization hotspots effectively. Moreover, by fostering a cultural shift within the organization, the entire team is now actively engaged in and understands the importance of resource optimization. This collective effort has created a company-wide focus on seeking opportunities to reduce the carbon footprint of their workloads. As a result, Holaluz is now well-positioned to measure and attribute both costs and emissions to their AWS workloads, empowering them to make data-driven decisions that align with their sustainability goals while maintaining operational efficiency.