We looked at how we were structuring our cloud services and reexamined our cloud environment using cloud analytics and reporting, data-driven insights, knowledge of vendor offerings, and an understanding of our compute environment. Our aim was to see where we could get the most value for Accenture, focusing our efforts on speed, sustainability and spend.
Our analysis led us to identify six key levers that enable us to optimize our cloud operations. These levers make operations leaner, smarter, greener, and continuously realize value.
They include:
- Application modernization – We are re-platforming legacy applications to serverless technology and cloud-native architecture while maintaining application functionality, increasing operational efficiency and cost savings.
- Right sizing and true sizing – We looked at how much compute power we needed, when we needed it and most importantly, when we didn’t need it. We measured instance usage and performance to understand compute needs, correctly match workloads and avoid overprovisioning. Doing this unlocked powerful savings that significantly reduces our operational costs and carbon emissions impact.
- Instance payment strategy – By understanding our vendor offerings and Accenture’s true compute needs, we found the most efficient pay structure that led to more informed decisions about how we pay for cloud services. To find the most efficient payment structure, we leverage spot and reserved instances for interruptible and predictable workloads.
- Instance upgrades – We run instance upgrades regularly that allow us to stay current, run faster and more efficiently at a lower cost. This ongoing upgrade process also keeps instances on the latest verified version, taking advantage of new releases from cloud providers.
- Environment rationalization – To prevent waste, underutilization and redundancy, we encourage teams to leverage the cloud’s horizontal scaling technology, automatically matching compute power based on usage. We also defined stricter standards and created governance structures for environment sizes and types. These efforts also include working in partnership with cloud providers who are transitioning to use renewable energy to power their data centers.
- Smart scheduling – We established automated, flexible shutdown schedules based on usage patterns that turn off servers in environments that are not in use, reducing costs and idle resource consumption. This effectively gives teams a way to “turn off the lights” at the end of the day and enables Accenture to operate in a greener way.
CO2 Calculator prototype
An innovative aspect of our program was the development of a CO2 Calculator prototype to produce detailed, data-driven measures of the environmental impact of our cloud environment. It’s based on a series of formulas to measure an estimation of Accenture’s public cloud energy consumption. This information enables our application teams with ways to lower energy usage and estimated carbon emissions.
Using algorithms, the CO2 Calculator analyzes billing and consumption data across Accenture’s global IT public cloud environments, and in turn, estimates energy and CO2e (e = equivalent) emissions, based on how frequently our server inventory remains powered on for use, as well as what size servers we choose to run within the environment.
This estimation results in more rapid optimization. By being able to make reasonable calculations as to the environmental footprint of a specific asset, like a cloud virtual machine, we can show how responsible and thoughtful management of even that one, individual asset can make a difference to the environment. The CO2 Calculator prototype creates a continual cycle by which stakeholders can consume the data and allow it to drive better business and environmental decisions.
A new way of working
Implementing these optimization efforts challenged long-standing assumptions within our global IT community about acceptable or recommended ways to operate one’s cloud environment and to shift to embracing a new mindset of thinking from a sustainability perspective. Our project team was able to incent a new way of working through data-driven guidance to understand the current baseline of cloud utilization and how much inefficiency existed.
We then focused on rethinking cloud consumption. Instead of basing consumption on risk mitigation, we changed the focus to be around fine-tuning consumption based on actual needs. We supported the application teams in being able to show total value back to their customers. We did this by showing how they are contributing to Accenture’s goal of net-zero carbon emissions as well as gaining the security that their decisions are being made based on intelligent data, not simply a cost savings discussion.
The more our program progressed and demonstrated that the guidance was not adversely impacting any of the business, the less people resisted and the more they came on board with the new way of managing cloud consumption.
