Calculating the Minimum Bounds of Energy Consumption for Cloud Networks

Abstract

Abstract-This paper is aiming at facilitating the energyefficient operation of an integrated optical network and IT infrastructure. In this context we propose an energy-efficient routing algorithm for provisioning of IT services that originate from specific source sites and which need to be executed by suitable IT resources (e.g. data centers). The routing approach followed is anycast, since the requirement for the IT services is the delivery of results, while the exact location of the execution of the job can be freely chosen. In this scenario, energy efficiency is achieved by identifying the least energy consuming IT and network resources required to support the services, enabling the switching off of any unused network and IT resources. Our results show significant energy savings that can reach up to 55% compared to energy-unaware schemes, depending on the granularity with which a data center is able to switch on/off servers.

Energy-Efficient Resource-Provisioning Algorithms for Optical Clouds

“…Depending on the scenario, the new integrated infrastructure can improve energy efficiency up to 45% and the EE resource orchestration up to 10%. Another attempt to define a comprehensive energy model where network and IT resources are treated in an integrated way has been examined in our earlier work [8]: it addresses the energy-efficient operation of integrated network and IT infrastructures in the context of cloud computing in an offline scenario. There, we proposed energy-efficient routing and IT allocation algorithms using MILP, by allowing switching-off several IT and networking elements and by exploiting the anycast principle.…”

“…On the other hand, pure network-energy minimization allows energy savings of the order of 1% to 2% of the total energy budget compared to shortest path routing (i.e., energyunaware). Although [20] and [8] indicate that treating network and IT resources jointly allows for energy optimization, their approaches are difficult to adopt in real settings since they suffer from scalability issues and cannot produce results in a reasonable time frame. Therefore, we extend this earlier work in two ways: (i) we update the energy model to include energyefficient cooling units (In-Row Cooling) and (ii) we tackle the problem in an online scenario to obtain results in a faster time frame.…”

“…This paper presents a heuristic that for a given request finds (i) an IT end point to process the request (the scheduling problem) and (ii) a route from the requesting source to that IT end point in the optical network (the routing problem). Requests arrive sequentially and we are solving the online routing problem, as opposed to the offline version (e.g., [8]), which has an a priori known request vector, expressing for each source the number of requests which need to be served. Our algorithm minimizes energy consumption by either trying to share as much active resources as possible (avoiding a startup cost for each newly activated resource) or by allowing switching-off idle resources.…”

Stochastic Energy Efficient Cloud Service Provisioning Deploying Renewable Energy Sources

“…This work is aiming at extending existing energy-aware service provisioning schemes ( [49]- [50]) that do not consider the stochastic nature of services, inherent in elastic cloud, or the unpredictable and non-steady power inputs of alternative energy sources. This is an important consideration, as it may lead to inefficient utilization of network and computing resources and increase in cost and CO2 emissions.…”

NCP+: An integrated network and IT control plane for cloud computing

“…This model amounts to anycast routing, solving the problem of selecting a route to one target destination chosen from a set of nodes, as opposed to unicast where the sourcedestination pair is known in advance. Anycast has been shown to be beneficial for the overall network performance, either in terms of network survivability [5,6], impairment avoidance [7], energy minimization [8,9] or blocking probability reduction [10]. Consequently, the control mechanism of an integrated network and IT infrastructure to select both the data center (i.e., the IT end point, without initially specifying its location) and the network resources (the path to the chosen IT end point) becomes critical for guaranteeing an efficient operation of the entire infrastructure.…”