Understanding the necessity of frugal computing is essential due to the finite nature of computational resources and its environmental impacts.
Scopes
Scope 1: Direct GHG emissions
- Covers direct emissions from owned or controlled sources.
- Occur from sources that are owned or controlled by the company, for example, emissions from combustion in owned or controlled boilers, furnaces, vehicles, etc.; emissions from chemical production in owned or controlled process equipment.
- Direct CO2 emissions from the combustion of biomass shall not be included in scope 1 but reported separately.
- GHG emissions not covered by the Kyoto Protocol, e.g. CFCs, NOx, etc. shall not be included in scope 1 but may be reported separately.
Scope 2: Electricity indirect GHG emissions
- By using the energy, an organisation is indirectly responsible for the release of these GHG emissions.
- Accounts for GHG emissions from the generation of purchased electricity consumed by the company.
- Purchased electricity is defined as electricity that is purchased or otherwise brought into the organizational boundary of the company.
- Scope 2 emissions physically occur at the facility where electricity is generated.
Scope 3: Other indirect GHG emissions
- Includes all other indirect emissions that occur in the upstream and downstream activities of an organisation.
- These emissions are a consequence of the activities of the company, but occur from sources not owned or controlled by the company.
- Some examples of scope 3 activities are extraction and production of purchased materials; transportation of purchased fuels; and use of sold products and services.
Scope 4: Avoided Emissions
- Avoided emissions are emission reductions that occur outside of a product’s life cycle or value chain, but as a result of the use of that product .
- Examples of products (goods and services) that avoid emissions include low-temperature detergents, fuelsaving tires, energy-efficient ball-bearings, and teleconferencing services.
- Other terms used to describe avoided emissions include climate positive, net-positive accounting, and scope 4.
Server footprint calculation
Total CO2 = embodied CO2 + CO2 from use This is equal to: (PUE * lifetime * electricity consumption * electricity carbon intensity) + embodied carbon
- Embodied CO2 = sum of emissions from manufacturing of all parts.
- Accurate figures are hard to find, but a good model is provided by Boavizta.
Assuming a server in a data centre
- server CO2 from use = lifetime * electricity consumption * electricity carbon intensity
- But data centre has overhead, mostly cooling
- Expressed as PUE, Power Usage Effectiveness, a number > 1
Total CO2 from use = PUE* server CO2 from use
Example: Dell PowerEdge R7515
- Embodied carbon 1400 kgCO2e
- Electricity consumption 415 kWh/year (medium load)
- Carbon intensity: UK 182 gCO2e/kWh; Glasgow 30 gCO2e/kWh
- 4 years lifetime
- PUE 1.5 (quite good)
Total: UK 1853 kgCO2e; Glasgow 1475 kgCO2e
Longer-term factors
- Energy efficiency still increases, about 1.2x per year.
- Carbon intensity decreases fast
- Every replacement increases the embodied carbon
- What is the optimal lifetime of the server?
Finite Computational Resources
- Historical Growth: Since the 1970s, there has been a significant rise in the usage of computational resources, with performance per Watt growing exponentially as per Koomey’s law.
- Limitations of Growth: Despite advancements, the finite nature of resources means that they cannot support indefinite expansion. Existing laws like Moore’s and Koomey’s cannot solely meet the increasing demands.
Environmental Impact
- Carbon Footprint: Computing devices not only consume significant energy but also contribute to a substantial carbon footprint due to their production and operational demands.
- Hardware Lifespan: The rapid obsolescence dictated by Moore’s Law leads to shorter hardware lifespans, exacerbating environmental impacts.
Frugal Computing
- Resource Conservation: There’s a critical need to acknowledge the limited availability of computational resources and to use them judiciously.
- Energy Efficiency: Professionals in the computing field must focus on maximizing energy efficiency to minimize environmental harm.
- Extending Device Lifespan: Prolonging the operational life of computing devices using existing technologies is vital for sustainability.
The Scale of the Problem: Meeting Climate Targets
- Climate Targets: To restrict global warming to below 1.5°C by 2040, global emissions need to be reduced drastically from 57 to 13 gigatonnes CO2 equivalent per year.
- Electricity Emissions: Current electricity emissions stand around 10 GtCO2e, with most electricity still being produced from fossil fuels.
Challenges with Renewable Energy and Carbon Capture
- Slow Deployment: The rollout of renewable and nuclear energy sources is lagging, with nuclear facilities taking up to 20 years to construct.
- Carbon Capture Issues: Carbon Capture and Storage (CCS) technologies face challenges such as high energy consumption and unquantified risks.
Limitations of Carbon Offsetting
- Finite Absorption Capacity: Terrestrial ecosystems can only absorb a limited amount of carbon, insufficient to significantly offset ongoing emissions.
Imperative of Emission Reduction
- Reduction Necessity: Achieving necessary CO2 level reductions will require a decrease in both energy consumption and goods production.
The Carbon Cost of Computing
- Growing Emissions: By 2040, computing is expected to account for 14% of total emissions, a significant increase from current levels.
- Production vs. Operation: Emissions from manufacturing computing devices are projected to exceed those from their operation significantly by 2040.
Life Cycle of Hardware Emissions
- Manufacturing: Includes emissions from mining, transportation, and manufacturing processes like chip production.
- Usage: Encompasses emissions from infrastructure operations, including electricity consumption and cooling.
- Disposal: Involves emissions from recycling, refurbishing, and landfill processes.
Full System Breakdown (Server to Screen)
- End User Devices: PCs, phones, tablets, and IoT devices all contribute to emissions through their usage and production.
- Data Center Infrastructure: Involves servers, network infrastructure, and cooling systems, all of which are significant energy consumers.
By understanding the comprehensive impacts of computing on sustainability, the need for frugal computing practices becomes evident, focusing on efficiency, extended device lifespans, and the overarching goal of reducing environmental impacts.