By Suraj Nair, Director – Technology, COE Leader, IOT and Telematics, Quest Global
Mass awareness of the negative effects of human activities on climate change has resulted in consumers becoming conscious of the impact of products that they buy and use on the environment. In addition, there has been intense pressure from policy makers too, to ensure organisations are public around their adherence to sustainability goals and the products they manufacture have minimal or zero impact to environment, economy, and society.
As more companies sign up pledges to become net carbon negative by 2050, this would mean that the whole organisation (including IT) optimizes and accounts for actions towards achieving sustainability goals – even for digital initiatives.
Adopting the cloud
While adoption of cloud technology in enterprises for their applications and products, has been primarily a business and technology decision, cloud adoption is also seen as a lever to reducing indirect emission footprint of data centers – data centers not always being receptive to dynamic scale needs (seasonal demands in e-commerce for example) and hence less opportunities towards managing energy consumption. Cloud vendors provide platform services that enable need based dynamic resource scale and workload consolidation to reduce server volume requirements. Cloud adoption works favorably to an organisation’s GHG protocol emission accounting needs (with emission contributions on running cloud workloads being classified under GHG protocol Scope 3)
Cloud vendors make the right moves
Cloud vendors today have adopted strategies towards achieving net zero. These include supporting digital initiatives in global renewable energy projects, water stewardship, adoption of renewable energy to power their data centers (from the grid or private deployments), use of low carbon concrete in data center construction and more. Also, cloud vendors are investing in building custom hardware that run their servers to improve data center efficiencies– AWS’ Graviton being an example.
Sustainability ‘in’ the cloud and ‘of’ the cloud
Like the principle of achieving security in the cloud, there is a clear segregation of responsibility in optimising for sustainability in the cloud. Cloud vendors are responsible for sustainability of the cloud (focus here being on building sustainable data center ecosystem, infrastructure, and platform services on top), while cloud customers (solution developers) are responsible for optimising their workloads for sustainability in the cloud.
Sustainable workloads on the cloud
Regulatory bodies around the world are increasingly incorporating the GHG Protocol into their environmental regulations, making it mandatory for organisations to track and disclose their emissions. As a result, “Sustainability” as a non-functional quality attribute requirement for cloud solutions is rapidly capturing the same amount of focus as other non-functional attributes (performance, scalability, portability, performance) in software architecture.
1. Identifying the right cloud – Customers hence need to understand how cloud vendors are helping achieve sustainability ‘of’ the cloud and their alignment, investments, and progress to net zero (or negative) goals currently and in future. For example, datacenter spread in regions that source renewable energy in the grid is one such pointer. Commitment towards water stewardship is another.
2. Optimising digital cloud solutions– The practice of sustainability in the case of cloud workloads includes understanding the impact of services used, measuring impacts through the workload lifecycle, and using architectural best practices to minimise these impacts. Cloud solution architects and developers are responsible for making the right architectural choices to optimise “in” the cloud. This would include, among others, choosing the right type of compute instances (ARM vs x86 based general purpose compute), right managed cloud platform services, implementing data patterns to reduce space consumption, optimising algorithms, designing dynamic scale, strategies to level out peak loads where possible etc.
While grooming non-functional requirements, it is essential to continuously optimise requirements (keep end solution cost and impact on sustainability in mind) and yet be meaningful to customer and end consumer requirements (An example would be to relax availability metrics for non-essential needs and automate, rather than waste resources on standby for failure).
Leveraging the cloud for embedded application development and deployment
A related indirect example of how companies are leveraging the cloud to optimise production of development hardware units and hence control waste and emissions, is through the concept of Cloud based Virtual Development Environments for embedded applications. While developing embedded applications has been quite hardware intensive, the recent proliferation of ARM based compute instances (that consume ~50% less energy) in the cloud have enabled companies to enable development and deployment of embedded applications onto ARM instances that emulate the actual hardware. So, developers can develop and test functionality in the cloud to ensure functional correctness, without having to procure individual development hardware units for most of the development lifecycle. Considering the rising volume of embedded applications (the concept of SDV for automotive being an example), a virtual development and deployment environment on the cloud enables distributed development and eliminates dependency on machines and hardware to be always available. This minimises the need to manufacture and procure hardware boards for developers until it is absolutely needed (shift left development, shift right manufacturing), thus reducing unnecessary production and potential waste.
With the focus on sustainability, cloud vendors and other players provide solutions that enable cloud solution providers to monitor the carbon footprint post profiling of cloud projects – like for example the Carbon Footprint dashboard from GCP. In general, one would notice that decisions that optimise workloads for cost indirectly optimise for sustainability too.
