When it comes to a responsible digital policy, one of the biggest challenges is to accurately measure your environmental footprint. This is required in order to take appropriate action. Measuring your footprint allows you to identify the biggest contributors to your total environmental digital cost.
You cannot set digital Green House Gases (GHG ) targets without first implementing indicators and a measurement tool. In the current context, where much of the IT is outsourced, how do you go about this? How does FinOps provide reliable solutions?
Measuring as a starting point
Tackling the problem of accurate measuring is not simple. Digital workflows are increasingly focused on outsourcing and the public cloud (IaaS, PaaS and SaaS). Unlike traditional information systems, the scope of today’s system is no longer limited to your on-prem server rooms and the computer equipment they contain. The services are provided virtually, which means you do not know what physical hardware is really being used and should be included in your measurements. In addition, the data on the amount of energy consumed by this equipment is not readily available. The metrics provided by the cloud provider are used to deduce the usage and based on modelling assumptions, can be translated into CO2-equivalent emissions. Therefore, measurements are dependent on the information disclosed by service providers, such as the environmental impact of their hardware, actual consumption, energy mix, etc.
This doesn’t mean you should not leverage the benefits of the cloud. Indeed, pooling resources can enable the public cloud to achieve a higher level of technical performance thanks to technological advances. By its very nature, the practice allows organisations to share and cut down on costs.
FinOps best practices can lower environmental costs
This is where FinOps really comes in to play. FinOps refers to all the IT operations that relate to managing usage costs in the public cloud. One of the biggest issues is optimising the utilisation of resources to lower financial costs to what is strictly necessary. This focus on reduction in financial costs can be easily transposed to environmental costs; however, not all FinOps practices follow this pattern. That is why we must compare their impacts on energy consumption to measure differences based on the deployment of FinOps actions.
Here are some examples of practices that can help to reduce the carbon footprint of your cloud services:
- Decommission virtual machines. Removing any unused machine will reduce the environmental impact and remove unnecessary costs. This can be accomplished when the supplier pools and reallocates the resources to another client.
- Reduce allocated resources and use efficient templates. In line with the previous point, remove any unallocated resource from the scope, whether it is a computing, network or storage resource. Reducing allocated resources an also prompt teams to reconsider their usage and innovate to work within a more restrictive framework. In particular, recent cloud templates make more efficient use of resources and provide for better service.
- Optimise applications and challenge business needs: This tactic is frequently overlooked because it requires investing in a code review and erasing the technical debt. Nevertheless, many best practices related to simplifying code and applications architecture are also best practices in eco-design. Along the same lines, proper scaling to match a business need is not systematically redundant and highly
- Free up storage space or delete unused applications: This action can reduce your impact because it also comes under the heading of reducing resources. It is important to note, however, the reduction is less significant here because the cloud is usually already optimised not to make these resources highly available, which means they are not necessarily powered.
The downside of the cloud service provider business model
Beyond reducing usage, some FinOps practices are at odds with the reduction in GHG:
- Use of reserved instances: Although reserving instances lowers financial cost thanks to the discount granted by the provider in exchange for a use commitment, it does not reduce resources. The opposite effect could even occur, because employees tend to want to use all the resources made available to them without questioning their needs ahead of time.
- Use of Spot instance: This practice does enable the pooling of unused resources belonging to the provider's other clients, but it also means that the instances are not switched off. Depending on the chosen scope, the result could be increased emissions, yet optimised costs.
Multiple methods to calculate your CO2 equivalent
In 2021, Wavestone worked with its clients on a calculation methodology tied to the tool and open source data from Cloud Carbon Footprint. The cloud leaders (AWS, Azure and Google) all developed their own products in parallel. As the calculations of each tool are based on diverse hypotheses, none can be used to compare different cloud service providers to one another.
The results are merely orders of magnitude that can serve as a benchmark to chart a course toward reduction. Because all the tools and modeling assumptions are rapidly evolving, it is important to acknowledge and embrace the need for a continuous improvement mindset, such as:
- Advance gradually towards covering a comprehensive scope
- Improve modelling over time by using emissions factors taken from external databases
From FinOps to GreenOps: an undeniable challenge
While FinOps is an entry point to steer your carbon transition, you must take the time to think about any positive or negative correlation between the financial and environmental cost variables. Many measurement tools have been introduced and are constantly improving; therefore it is crucial that they lead to relevant and personalised multi-cloud recommendations that address the sustainability issues that are top of everyone’s mind. The transparency about modeling assumptions used also fundamental.
Today, the biggest source of cost savings achieved by our clients through FinOps is the use of reserved instances. The environmental merits of this option are questionable, so, the next challenge for cloud providers is to update their recommendation tools to take into account alternatives to work around these reserved instances in order to stay as close as possible to the cost-per-use model.
Finally, the aim of FinOps, and “GreenOps” in to achieve cost saving targets. Businesses are still faced with a battle to achieve reduced emissions and costs up front, rather than recuperating on existing infrastructure. It is a level of maturity that requires a major cultural shift by all the stakeholders to overcome the illusion that a virtualised resource is an unlimited resource.