Setting up your data on cloud sustainably
Co-authors: Akilesh S. Rajeev John
In our first blog we introduced data and sustainability, building on which in our second blog we discussed the sustainability challenges one might face while setting up their data on cloud. Our current blog presents ways in which data could be setup on cloud sustainably.?
Sustainability Strategy
As sustainability becomes a necessity to be more responsible and as the need for sustainability reporting increases, businesses face increased pressure from consumers and stakeholders to be more transparent with their sustainability goals and initiatives. From informing purchasing decisions of consumers to investment decisions of stakeholders, the sustainability strategy of businesses has started to play a vital role. Gartner reports that 90% of organizations have increased their sustainability investments since the pandemic compared to 2017 levels, as Environmental, Social, and Governance (ESG) reporting has gained more attention.
A thorough sustainability strategy would help businesses be more responsible toward the environment, communities they work with, and would also shed positive light on their business. The strategy can vary based on the industry, size of operation, geography, among others. While designing a sustainability strategy it would be important to consider the entire supply chain of the organization – from the energy mix they use to the technologies they implement. They could also identify their different verticals’ environmental impact, ideate on measures to mitigate the negative impact, understand the timeline in which the measures would deliver results, and set KPIs to measure what success would look like and to keep track of the goal.
Cloud could boost the sustainability scores of businesses, as it could have various environmental benefits as discussed above. While moving to cloud or modernizing the data infrastructure, the major considerations have usually been the organization’s business and technology strategy, but it should also be in line with their sustainability strategy. This would help organizations make greener choices with the technology partners they choose and how they architect their technology. A sustainability strategy would also help in the long term by being aware of future emissions and how it compares with our targets.
Sustainability Evaluation
One aspect of the sustainability strategy would be of businesses’ technology. To understand their technology’s implication on sustainability they could start by evaluating its current state and define a future state to set realistic and achievable goals. A few parameters they could evaluate are the emission of their data centers, carbon footprint of their hardware, software design practices, energy consumption of their AL / ML models, sustainability measures of cloud service providers they are considering, their energy mix, and value realization by switching to greener options.
Data center’s emission – As we had discussed in the previous blogs, on-premise data centers are far less utilized and consume more energy. In a study titled “The Carbon Benefits of Cloud Computing: a Study of the Microsoft Cloud”, Microsoft considers the IT infrastructure’s entire lifecycle phases to measure their energy consumption and carbon emission. The phases they considered are raw material extraction and assembly, transportation, use, and end of life disposal. This comprehensive approach gives a more precise estimation of how much carbon a data center would emit, calculated from the individual equipment used. We could also take into consideration their cooling methods and if they cause any environmental impact (Ex – gallons water consumed per day if a water-based cooling system is used).
The source of their power could be identified through guarantees of origin (GoOs) or pacts such as the power purchase agreement (PPA).
Software design practices – Software development practices also influences the sustainability quotient for organizations. Bad design contributes to higher energy consumption, carbon emission, sub-optimal server utilization etc. To address these concerns and become more sustainable organizations should educate their developers on Green software engineering, which at the intersection of climate science, software, architecture and data center design, energy, and hardware enables them to be better informed and environment conscious in their work. Its core principles talk about building applications that are energy, carbon, and hardware efficient, uses energy with low carbon intensity, maximizes hardware’s efficiency, reduces amount of data and the distance it must travel along the network, incrementally optimization that improves carbon efficiency, and being carbon aware.
AI’s energy consumption – We saw in our previous blog how AI can be detrimental to the environment if it is not built responsibly. Organizations should assess their current practices in developing AI, identify the carbon emissions by their models, and educate their employees on sustainable AI practices. There are also tools available to evaluate carbon emission of ML models.
Aimee Van Wynsberghe introduces “Sustainable AI” movement in her paper titled “Sustainable AI: AI for sustainability and the sustainability of AI”, which is aimed at fostering change in the entire lifecycle of AI products – idea generation, training, re-tuning, implementation, and governance towards greater ecological integrity and social justice. Hence it is important to focus not just on the development and implementation of AI models but holistically on their entire lifecycle.
Organizations should also evaluate the value of the AI model vs its environmental impact and make a call accordingly on whether they would need the model. They could also use energy efficiency indicators for any model they develop thus helping to understand its impact when it gets reused.
Greener Tech Choices
Once the current state is evaluated and sustainability targets are defined, organizations must look for greener choices to move to. This could be any technology build, buy, or rent. For example, AWS allows their customers to understand their energy efficiency, cooling efficiency, energy mix, provides tools to track their carbon footprint, and performance efficiency. All major cloud providers are transparent in various aspects to their sustainability measures.
Choices are plenty while choosing a cloud provider and it can get confusing and difficult to identify the different sustainability parameters to evaluate them on. There are certain tools in the market that help assess carbon footprint of existing data centers and applications and facilitates comparison among cloud migration options and the benefits associated. Thorough research should be done by organizations to examine the suitability of the tools.
领英推荐
Optimizing Cloud for Carbon Emissions
Getting data on cloud is sustainable by itself, but how can one optimize their cloud usage to further reduce consumption and hence be more sustainable? Following are some ways of optimizing cloud utilization and hence reduce the carbon footprint. Cloud is designed to provision resources as and when required, so organizations could look at creating resources only when there’s a requirement and not have them idle or running. Similarly, they should also have periodic checks to ensure that idle or unused resources are deleted, deprovisioned, or reassigned. Elasticity of the cloud could also be leveraged by moving workloads to managed services, which scale as per need. While migrating cloud, legacy monolithic applications which are rigid with a larger carbon footprint could be refactored into microservices which are lighter and are easier to scale.
Positioning Data Centers
While choosing a region to store, process, or access data from, organizations usually consider the data regulations (sovereignty), cost of operations, latency, features offered, SLA, etc. Another important parameter that should be factored in while deciding on a region is the sustainability aspect.
Sustainability of a region could be evaluated by looking at factors such as the carbon free energy they use, the policies mandated by the local government on sustainability or energy consumption, its geography – colder regions would require lesser cooling and hence lesser water or coolant to offset the heat, per square foot area used to store the servers. These are some factors that would help narrow down choices while looking for sustainable regions.
Sustainability could also be factored into the other parameters. Data sovereignty, for example, dictates if an organization operating out of a region is allowed to store or process its data outside the home region and this may affect the ability to choose a region which is more sustainable, as data cannot leave the region it originates from. But these restrictions mostly apply to the PII data that is collected, so organizations could consider retaining the PII data at origin but move other data to more sustainable regions. For data sovereignty restrictions, the governing bodies and policy makes should bring in measures which would make data centers more sustainable in their regions. For example – they could incentivize use of green energy to power data centers or encourage innovation in the sustainability space to make data centers greener.
With other factors as well such as latency, SLA, and cost, they can evaluate on how fast they would need the data, how often they would need it, etc. and decide accordingly which parts of data should be stored in regions closer to the users and which parts could be stored in distant albeit more sustainable regions.
Value Realization
More than being environmentally friendly, sustainability choices could also provide monetary benefits. Businesses could also evaluate the monetary benefits their sustainability choices could bring in.
The Indian government, for instance, is planning to offer a 3% incentive on data center investments for the use of renewable energy where possible. Since several governments are moving toward greener initiatives, these incentives would provide the necessary push while benefiting businesses as well.
As customers are growing more environmentally friendly, they are factoring the same in their decision of their product / services. Businesses can hence reposition themselves as sustainable by emphasizing their sustainability goals, their (planned) actions to uplift affected communities, and how green their current operations are.
In our next blog we discuss what kind of insights one could derive from their data to enhance their sustainability quotient.
References
Previous Blogs:
This content is provided for general information purposes as the views are personal and is not intended to be used in place of consultation.
Managing Director, Data & AI | Leading Data & AI Transformation' at Accenture
2 年Nice blog Saurabh Pathak and Akilesh S. ! I look forward to the next one on enhacing sustainability quotient through data!