Balance your Bids before your Bits: The Economics of Geographic Load-Balancing

By routing workload to locations with cheaper electricity, geogra-phic load-balancing (GLB) has been shown a promising mecha-nism to cut down the electricity bill of geo-distributed datacenters operated by the same organization. Most existing studies on GLB assume that the use of GLB has no impact on electricity prices, even though GLB increases local electricity demand variation. In practice, however, electricity prices are determined by how supply and demand are dynamically balanced by local electricity utilities, and thus may as well be affected by GLB. In this paper, in order to understand and unleash GLB’s economic potential, we carry out a comprehensive study on how GLB interacts with electricity supply chains. In particular, we show that as GLB introduces extra un-certainty in local demand, utility companies may have to increase electricity prices to ensure certain profit margin in face of such de-mand uncertainty. Consequently, cloud service providers (CSP) do-ing GLB may end up getting minor cost reduction or even paying higher electricity bills than not doing GLB, as shown in our case study based on real-world traces. Then, motivated by the recent practice of large CSPs moving into electricity markets, we propose to allow CSPs to purchase electricity from markets through bro-kers. The advantage is that GLB no longer causes economic loss to utilities. Meanwhile, CSPs can still exploit their presence in mul-tiple geo-locations to achieve desirable electricity cost reduction. Our case study using real-world traces shows that the solution can save CSPs up to 12 % of the electricity cost