1. Economic and Environmental Impact of Optical Data Center Networks
Arash Deylamsalehi
Advisor: Vinod M. Vokkarane, Ph.D.
Information and Communication Technology (ICT) consumes about 1500 TWh of energy
annually. ICT is responsible for the release of 789 megatonnes of carbon dioxide emissions and
$90B of electricity expenditures. Both impacts have been increasing dramatically with the growing
size of digital universe. Energy consumption increase has been addressed by attempts at improving
energy efficiency during last decade. The increase in energy efficiency achieved by researches
started around 2005. Traditional practices assume that increasing energy efficiency is sufficient
for minimizing total overall energy consumption. Unfortunately, studies indicate that more energy
efficiency leads to less efficient consumption, and more actual energy consumption. Therefore,
ICT operators need to focus not simply on energy efficiency but increase efforts to reduce
emission, and the cost of electricity.
Today’s communication networks span cross-country or cross-continent regions. Thus the
network companies distribute their infrastructures in different geographical locations worldwide
to improve their reliability and quality of service. Such large and often diverse areas may be served
by multiple competing power markets throughout; the electricity price and fuel mix differ from
site to site at any given time, and from time to time at any given site. Moreover, data replication
allows content to be available from several optical data centers that may be located in different
geographical locations. Such destination-agnostic routing can be defined as an anycast request in
order to sufficiently reach one of the data replicas. Thus shifting a partial amount of electricity
consumption from a given location to other alternative locations leads to a significant potential for
reducing electricity cost and/or emissions of networks. Energy-aware routing is capable of
minimizing the electricity cost and emissions in the core and at the network edge, including optical
data centers. However, although some types of conventional power plants (particularly coal power
plants) are cheap sources of electricity, they are also the world’s top contributing sources of carbon
dioxide emissions, and the primary cause of global warming. Thus, reduction of electricity cost
and reduction of carbon dioxide emissions may be distinct goals in opposition to one another.
Considering the electricity cost and emissions of power markets in a combined manner may yield
superior green electricity cost saving or economic green networking in the optical networks, to
address both economic and environmental concerns about ICT as a joint concern.
The purpose of this thesis is the development of novel energy-aware routing paradigms as an
interdisciplinary project between power grids and ICT, to make optical networks truly economic
and ecological without compromising their praised qualities of reliability and efficiency.
Approaches include the development of novel analytical models and frameworks to quantify
economic and environmental impacts of network operations, routing algorithms to reduce the
electricity expenditures and emissions of network operations, and solutions for implementing
green electricity cost saving and economic green networking.