Online virtual machine migration for renewable energy usagemaximization in geographically distributed cloud data centers

摘要

Energy consumption andits associated costs represent a huge partof cloud providers’ operationalcosts. In this study,weexplorehowmuchenergy cost savings canbemade knowing the future levelof renewable energy (solar/wind) available in data centers. Since renewable energy sources haveintermittent nature, we take advantage of migrating virtual machines to the nearby data centerswith excess renewable energy. In particular, we first devise an optimal offline algorithm withfull future knowledge of renewable level in the system. Since in practice, accessing long-termand exact future knowledge of renewable energy level is not feasible, we propose two onlinedeterministic algorithms, one with no future knowledge called deterministic and one with limitedknowledge of the future renewable availability called future-aware.We show that the deterministicand future-aware algorithms are 1 + 1∕s and 1 + 1∕s − ğœ”∕s.Tm competitive in comparison tothe optimal offline algorithm, respectively, where s is the network to the brown energy cost, 𜔠isthe look-ahead window-size, and Tm is themigration time. The effectiveness of the proposed algorithmsis analyzed through extensive simulation studies using real-world traces of meteorologicaldata and Google cluster workload.