SUSTAINABLEELECCOST: OPTIMIZATION IN GREEN DATA CENTERS FOR EMERGING MARKETS

Abstract

In response to the escalating demand for computing resources in modern datacenters,
it has become imperative to address this challenge efficiently while concurrently
minimizing energy consumption and costs. Recent advancements in power market
operations have introduced innovative demand-response programs, where electricity
consumers, including datacenters, can adjust their power usage in response to
provider requests, thereby optimizing monetary expenses. Among these programs,
regulation service (RS) reserves stand out for datacenters due to their lucrative credit
gain potential and the inherent flexibility of datacenters in regulating power
consumption. Consequently, it is crucial to devise effective bidding strategies for
datacenters to actively participate in emerging power markets, coupled with dynamic
power management policies that are attuned to real-time power market requirements.
In this paper, we introduce ECOGreen, a comprehensive strategy designed to
holistically optimize the datacenter RS problem and virtual machine (VM) allocation,
ensuring compliance with hour-ahead power market constraints. ECOGreen takes into
account the presence of electrical energy storage (EES) and renewable energy sources.
We employ a fast analytical approach to determine optimal power and reserve bidding
values, along with the optimal number of active servers, providing practical and
effective solutions. Furthermore, we present an online adaptive policy within,ECOGreen that modulates datacenter power consumption. This is achieved by
controlling VMs CPU resource limits, efficiently utilizing demand-side EES and
renewable power sources, all while upholding quality-of-service (QoS) constraints.
Our results underscore the effectiveness of ECOGreen, showcasing its ability to
contribute an average of 76 percent of the datacenter power consumption as reserves to
the market. This remarkable achievement is attributed to ECOGreen's adept operation
on renewable sources and EES. Notably, ECOGreen achieves up to a 71 percent
reduction in electricity costs compared to other state-of-the-art datacenter electricity
cost minimization techniques participating in the power market. This underscores the
significance of ECOGreen in achieving a harmonious balance between optimizing
datacenter operations, meeting power market demands, and ensuring cost efficiency.