How does cloud computing energy consumption compare with conventional computing? A new research report can help to improve strategic decision making on sustainability and cloud computing.
Most studies of energy consumption in cloud computing have focused only on the energy consumed in the data center. Although research showed that the energy consumption of data networks is something that cannot be neglected (see blog ), up to now the role of the network is almost entirely overlooked in the discussion of how green is cloud computing.
Recently a very interesting report on this topic came available and some of the results were presented at a seminar of Green IT Amsterdam. The research was conducted by SURF. SURF is the collaborative organisation for ICT in higher education and research in The Netherlands.
The report not only discus the issue of energy consumption of data networks. It take the discussion to another level by considering two different scenario’s:
- The ‘Bit-to-Energy’ scenario; with data being moved to ‘greener’ remote datacenters
- The ‘Energy-to-Bits’ scenario; with ‘greener’ energy being moved to the datacenter where the data resides.
Based on those two scenario’s the report tries to answer the following research questions:
- What are the sustainability effects of data transport over the data network? How much energy is required and what is the CO2 footprint?
- What are the sustainability effects of energy transport? When is it suitable to acquire green energy from elsewhere?
As stated in the report, when looking for the ‘greenest’ way to perform computationally intensive tasks, a user may have different options:
- Perform all computing locally, powered by locally produced energy.
- Perform all computing at a remote location, powered by energy that is produced sustainably at the remote location.
- Perform all computing locally, powered by energy that is produced sustainably at a remote location.
So for those enterprises that takes sustainability seriously, one of the major trades-offs to be made within an enterprise cloud strategy is how to distribute data storage and data processing across the cloud.
Basic energy models were developed for each scenario, the ‘Bit-to-Energy’ and the ‘Energy-to-Bits’ scenario, with three different use cases (CPU-intensive, Interactive and
Data storage). The first analysis showed that:
- The energy required to transport the data and the energy required for energy transport can be considerable and cannot be neglected in evaluating the overall level of sustainability in the various use cases.
- Application features as data processing times or required storage play a significant role in the final outcomes.
- One important conclusion is that in many data scenarios where the local data centre can import cleaner energy from elsewhere, the best course of action is to keep the data local and perform calculations locally.
Based on these models a website calculator is made available.
The report “Transporting Bits or Transporting Energy, Does it matter?” is available here.