The environmental footprint of a distributed cloud storage

Every time we access a file on the Cloud, a chain of routing servers is activated to transfer the payload from the storage facility to the user’s device, and back. The data center, being densely packed and highly-performant, is hardly optimized in terms of energy consumption and environmental impact. The ICT (information communication technologies) ecosystem is estimated to be responsible for the 10% of the full worldwide energy consumption equivalent to Germany and Japan total energy demand. However, being a fast-inflating and almost unregulated market, the environmental footprint caused by data storage and transfer on the internet shows no signs of slowing down. In this paper, we analyze a reversal paradigm for cloud storage (implemented by cubbit.io) in which data are stored and distributed over a network of p2p-interacting single board computers. We compare Cubbit to the traditional server-based solution in terms of environmental footprint and power/usage efficiency. We demonstrate how a distributed architecture is beneficial for impact reduction on both sides of communication and storage. First, being virtualized and distributed upon small single-board computers, storage is far more efficient than data centers racks and does not need any additional cooling. Secondly, the distributed architecture can leverage proximity of files to the user (e.g. within the same city) to minimize the consumption due to the powered route from the virtualized storage facility to the user’s device. We hereby quantify both these effects using the public internet model published by Baliga et al. (Proceedings of the IEEE 99.1, 2011), and compare the estimation with the same model applied to a server-based cloud storage (e.g. Dropbox). Results show that a remarkable reduction in terms of energy consumption is obtained on both storage (down to 8% of consumptions) and transfers (c.a. 50% of consumptions). 1 ar X iv :1 80 3. 06 97 3v 1 [ cs .D C ] 1 9 M ar 2 01 8