Cost as a proxy for carbon – the inconvenient truth
Many technology-related articles and major cloud service providers promote the idea that ‘Cost is a proxy for carbon emissions’. This suggests that optimising and reducing your cloud costs can lead to increased sustainability. But is this entirely true?
While there may be some truth to the statement, it is critical that the limitations to this approach are well understood.
So let's jump in. Cloud bills are calculated using an extremely simple formula: Cost = Rate* Usage. Let's see how each of these may impact the sustainability of your service.
Rate
Rate may be calculated in many ways depending on the service, and there are strategies that can be implemented to reduce this rate. These strategies include reservations, savings plans, committed use discounts and special pricing which you may have negotiated with your provider. None of these strategies will have an impact on the sustainability of your service as they are purely billing concepts. Indeed, some of them may drive negative behaviours where in order to avail of a discount or special pricing you do not perform an optimisation as it may impact this or cause you to be financially penalised if you don’t achieve a committed level of spend required to achieve a discount.
The problem with rate optimisations
We have already seen this tension of pricing acting as a disincentive for optimising for sustainability with some of our customers. For one UK government customer we found that optimising their cloud platform we would mean to powering off or deleting instances that we had already purchased reservations for. These unused reservations still incurred charges, potentially discouraging the customer from approving the optimisations. However, the customer was also focussed on the sustainability improvements and so in this case we were able to reduce costs by 36% whilst also reducing emissions by 42%.
Usage
Usage is something that may be linked to the sustainability or emissions for your service. Strategies to optimise your cloud costs through usage optimisations are where the FinOps and GreenOps disciplines intersect. However, the meters used to determine cost are set by the cloud vendor and as such it may not correlate to emissions. You may also find that the rate changes within a service depending on your usage, for example storage costs could reduce depending on volume of data stored. This will have an impact on the cost of your service, but not necessarily the related emissions.
In Infrastructure as a Service (IaaS), billing is time-based, not work or utilisation based. Using cost as a proxy assumes full utilisation, often not the case. Resource utilisation and the electricity and cooling they require correlate directly with emissions. Hence, cost as a proxy can distort the sustainability impact of optimisations.
But it's even more complicated than that...
A free service will still be responsible for creating emissions to provide that service. For many small use cases the free tiers are suitable. Does this mean that the service is greener or has zero emissions? What if the volume of traffic to my service falls such that I no longer breach thresholds where the cloud service generates cost?
What about the foundational services that you require in order to build out your own cloud services? Services such as IAM, or monitoring are all free, but they require hardware to operate and therefore energy and cooling. Cost will not take these into consideration but you will need to account for these in any reporting that you are required to submit.
A final example of this disconnect is that cost for cloud resources can differ across regions. This may not have a direct correlation to the sustainability of the service in the regions. It may be that lower pricing may be using more carbon intense energy mixes. Therefore, moving to a lower cost region may have a negative impact on the emissions of your service. Alternatively, regions with a cleaner energy mix may have higher prices for some services and as such your costs would increase were you to move to these regions. This would inaccurately penalise you for moving workloads to a greener region.
As you can see, using cost as a proxy for sustainability does not always track, and may actually be totally inaccurate. In part 2 we will outline an alternative approach, as well as where you may be able to use cost as a data point in your calculations.