Charting a course toward more sustainable blockchain solutions

Energy consumption, climate impact and greenhouse gas (GHG) effects are challenges that every responsible business is examining and trying to minimize.

At the core of the sustainability conundrum

At Atos, we have made climate sustainability a top priority and committed to a net-zero target (NZT). Around the same time, for the last seven years, blockchain technology has been gaining traction — with the potential to revolutionize the way we verify transactions, exchange value and transfer ownership. However, it has also been facing close scrutiny over the energy-intensive operating model of Bitcoin, the most prominent implementation of blockchain.

Depending on certain conditions and parameters, the Bitcoin network can consume between 72 and 324 terawatt hours (TWH) each year. By way of comparison, the estimated total electricity consumption of all Canadian households was 173 TWH during 2018.

The recently concluded Paris Upgrade of Ethereum Mainnet has been in the news because of its capability to change the way transactions will be approved on the Ethereum blockchain. This is expected to move Ethereum away from the older, more carbon-intense Bitcoin blockchain. However, it is yet to establish its value by demonstrating a real greenhouse gas emission reduction.

The key components of blockchain architecture

When we look at blockchain or distributed ledger technology (DLT), we expect them to deliver the following key benefits:

• Trust-agnostic governance
• Provenance tracking
• Ultra-high resilience with no single point of failure
• Finality of transaction

These benefits are only possible if the technology can boast of security, scalability, and redundant processing in blockchain architecture — critical factors that influence the energy required to operationalize any blockchain.

We must state here that a blockchain DLT solution will usually be more complex than a well-architected centralized solution. So, we should only use blockchain if the use case justifies the inherent complexity. Those who follow this topic will already know that, while there are many studies focused on Bitcoin and proof of work (PoW) blockchains, very few have studied non-PoW blockchains.

The Atos Expert Community and its Special Interest Group on decarbonization have been following this space closely and conducting research to distinguish between the energy consumption of the Bitcoin blockchain and an enterprise blockchain implementation.

As part of the study¹, we evaluated Bitcoin’s energy consumption and created a calculator to understand its carbon footprint while we focused on private blockchain implementations of varying consensus size, node size and transaction throughput.

The study revealed that while the popular PoW consensus mechanism in public blockchains like Bitcoin is an energy guzzler, non-PoW blockchain solutions can be fine-tuned to have energy consumption comparable to that of traditional centralized digital solutions.

The next logical question is: Can we create a parameterized framework to compare the GHG footprint of a blockchain implementation with other blockchain or non-blockchain solutions?

To answer that question, we developed two models – a bottom-up model and an empirical model.

A deep dive into the bottom-up model

Let’s take a look at the bottom-up model first. In the bottom-up GHG estimation of a blockchain, we consider three computing components – the CPU processing, storage, and network requirements.

Bottom-up GHG estimation of a blockchain

Within the CPU, we estimate the CPU cycles needed to drive consensus, validate the transaction, and execute smart contracts or chain code on each computing node in the blockchain network. For storage energy estimation, we evaluate the size of the ledger and the size of the world state — which is a key value database to increase reporting efficiency at a lower cost. The third component is the network, and its energy needs can be estimated by the throughput in terms of blocks and transactions, and the average size of all transactions.

Here’s how these different components add up to provide the answers we need:

• Multiplying all these estimates by the energy consumption per unit of that component, we arrive at the total energy consumption per blockchain node.
• Multiplying the result by the number of validators required in the network, we can estimate the total energy consumption of the network.
• To calculate the total GHG emissions, we multiply the network’s consumption by the weighted average of the GHG footprint of the energy sources that power it.

Although the math seems rather simple, putting it into practice is a different story — because it’s extremely difficult to get an accurate estimate of each component’s energy consumption up-front. So, rather than trying to get an absolute number for the GHG estimation, it’s more instructive to look at a comparison between two blockchain solutions. For this, we created an empirical model to help decide for or against blockchain when trying to implement a specific solution.

Weighing in on the empirical model

The empirical model weighs various factors of a blockchain solution to calculate a relative GHG emissions score.

Empirical model: Factors contributing to blockchain’s GHG footprint

Here, the estimation depends on ten factors that can also be used to fine-tune the GHG footprint of the blockchain network as per an organization’s business needs.

1. Solution scope: Is it a public blockchain, or within an organization or consortium?

2. Type of blockchain consensus: While blockchain solutions can employ many consensus methods, proof of work (PoW) and proof of stake (PoS) dominate, with PoW being the most decentralized and secure. This also causes it to have the highest energy consumption (by a factor of 3). PoS is emerging as a leading method, but others are seeing increased adoption.

3. Size of the main network: How many network members and validators exist?

4. Layer-2 network extensibility: Can we extend the main network to a Layer-2 network?

5. Required network redundancy: How much redundancy is required for data and processing?

6. Average transaction complexity: What is the average transaction complexity? This business logic complexity is encapsulated in smart contracts. Apart from consensus, smart contracts are the key driver of the blockchain solution’s GHG footprint.

7. Average transaction size (MBs): What is the average transaction size?

8. Required throughput (per sec): How much transaction throughput do our business use cases require?

9. Hardware efficiency: How efficient is the hardware we are implementing our blockchain nodes on?

10. Energy source to network: Finally, but probably the most significant question: what energy source is powering our blockchain hardware?

In the empirical model, we parameterize and enumerate different values of these factors to derive an overall GHG score for a blockchain solution.

Embracing sustainable solutions in blockchain

If blockchain is to become the future of secure digital data, it must also prove its sustainability value. Only then will it be widely accepted and adopted by organizations looking for future-ready solutions and exponential business growth. Once we have estimated the scope of a blockchain solution, we can critically evaluate our choices and take the next step to fine-tune these parameters.

In a white paper to be published later this year, we will delve into this research on blockchain GHG emissions and share our findings and the models. Ultimately, our goal is to provide a guide that will help business leaders evaluate how to incorporate sustainable blockchain solutions into the enterprise technology landscape. Stay tuned!

¹ The paper on the research conducted by the Atos Expert Community (AEC) and its Special Interest Group (SIG) on energy consumptions of the Bitcoin blockchain and an enterprise blockchain implementation will be published shortly. Watch this space for more.

Topics

Essentials
Blockchain
Environment