The Impact of Decentralised Computing on Sustainability: is there more to the story?

It has been estimated that computers, data centres, and networks consume about 10% of the world's electricity. Technological innovations such as blockchain have been proclaimed to significantly increase this number if they were to be widely adopted. The Ethereum cryptocurrency is the most notorious example of the environmental risk these technologies represent. Unfortunately, this discussion has given decentralised computing a bad name in terms of sustainability. 

In order to understand why blockchain is accused of having a negative sustainability impact, there are two terms we need to remember: decentralised computing and decentralised control. Decentralised computing is computation done on multiple devices instead of a single, centralised one. Decentralised control, then, means that there isn’t a single system or device that sends out commands and controls the process. 

Blockchain is known for utilising decentralised computing and decentralised control to increase data security.  Decentralised control is used to prevent a single adversarial party from taking control of the network easily. This is especially relevant when dealing with financial transactions, such as in the case of Bitcoin or Ethereum. Computation happens on thousands of devices simultaneously, while each device has an identical copy of the data. In this system, all devices must check the data registered in their ledger and confirm that they got precisely the same result, before a transaction can be permanently added to the public ledger. Naturally, this amount of duplicated computing may be seen as “unnecessary” energy wastage: the same calculation runs thousands of times, and it does so for every single transaction. 

However, there is more to decentralised computing than blockchain technology. An example of another type of decentralised computing is edge computing. This method of analysing and managing data is used as an alternative to cloud computing, a method that has become increasingly popular over the past decade, or even more traditional methods such as on-premise data centres. The main difference between the technology used for blockchain and edge computing is that while blockchain technology decentralises computing and control, edge computing technology decentralises computing but maintains centralised control. 

What, exactly, does that mean? In cloud computing, all data is shipped from its original source to a single extensive database, to be computed together. Edge computing relies on the end devices, such as a users’ phone, to do the calculations locally and only shares some of the results across its network. Instead of duplicating the exact computation across all devices, each device calculates a small portion from the data available to it. Essentially, the overall system trusts that the data available to each device is, in fact, accurate.

Now the question is, is edge computing better for the environment than cloud computing? One may be tempted to conclude that cloud computing must be more efficient since you can install more efficient heat dissipation mechanisms when calculating data altogether. However, that is not the case. The reason for that has to do with people and the choices we tend to make when we need to decide between environmental savings and financial ones.  

Cloud computing environments tend to encourage people to use more complex models which, first of all, are easier to design and, secondly, maximise the more powerful hardware capability. However, as they are complex, they also consume more energy. Moreover, risk management leads to the data being duplicated within a single data centre as well as across multiple data centres, which again comes at an energy cost. Finally, there is the cost of data shipment. Each time information is sent out; there is an energy payment attached. Cloud computing relies on constant transfers, both to the main storage and to the duplicate data centres. 

Edge computing tends to be more energy efficient because most of the data never leaves the device, thus minimising the ‘shipping cost’. Additionally, the computational models must be extremely energy-efficient, as the device's battery life impact must be kept to a minimum. 

These are some of the reasons why we, at NumberEight, chose to rely on edge computing for our platform. On top of the above considerations, we were also able to improve the computation’s sustainability by using models that constantly evaluate how much data is needed and only collecting, processing and storing the minimum amount of data required, thus reducing calculation ‘wastage’. 

Finally, evaluating the actual sustainability impact of the three computational methods can be challenging as there are many variables to take into account. However, there are two principles you should keep in mind. The first is that shipping data is energy expensive, as the device’s wifi and/or cellular antenna needs to be activated, which consumes a lot of power. The second is the complexity of the model at hand: the more complex the model, the more energy it consumes. One of the advantages of edge computing is that its very engineering limitations force us to be more efficient, leading to more sustainable computing.