Section: New Results

Energy Efficiency of Large Scale Distributed Systems

Participants : Laurent Lefèvre, Daniel Balouek Thomert, Eddy Caron, Radu Carpa, Marcos Dias de Assunçao, Jean-Patrick Gelas, Olivier Glück, Jean-Christophe Mignot, Violaine Villebonnet.

Energy efficient Core Networks

This work [8] , [43] seeks to improve the energy efficiency of backbone networks by providing an intra-domain Software Defined Network (SDN) approach to selectively turn off a subset of links. To do this, we designed an energy-aware traffic engineering technique for reducing energy consumption in backbone networks. Energy-efficient traffic engineering was analysed in previous work, but none addressed implementation challenges of their solutions. We showed that ignoring to test the feasibility of techniques can lead to bad estimations and unstable solutions. We proposed the STREETE framework (SegmenT Routing based Energy Efficient Traffic Engineering) that represents an online method to switch some links off/on dynamically according to the network load. We have implemented a working prototype in the OMNET++ simulator. Networks are progressively using centralised architecture, and SDN is increasingly utilised in data centre networks. We believe that SDN may be extended to backbone networks. The implemented solution shows that SDN may also be a good means for reducing the energy consumption of network devices. Compared to previous work, in this work we used the SPRING protocol to improve the stability of energy-efficient traffic engineering solutions. To the best of our knowledge, this is the first work proposing the use of SPRING to improve the energy efficiency of backbone networks. The flexibility of this routing protocol is well suited to frequent route changes that happen when we switch links off and on. Moreover, this protocol can be easily applied to SDN solutions. Using simulations, we showed that as much as 44% of links can be switched off to save energy in real backbone networks. Even greedy techniques can easily approach the maximum reduction in the amount of energy consumed. In fact, the bottleneck in terms of energy efficiency in energy-aware traffic engineering is the connectivity constraint. We performed a stress test of our solution under rapidly increasing traffic patterns and showed that more work must be done in the domain of switching links back on: a field which has received little attention from the research community.

Energy proportionality in HPC systems

Energy savings are among the most important topics concerning Cloud and HPC infrastructures nowadays. Servers consume a large amount of energy, even when their computing power is not fully utilized. These static costs represent quite a concern, mostly because many datacenter managers are over-provisioning their infrastructures compared to the actual needs. This results in a high part of wasted power consumption. In this work [19] , [47] , we proposed the BML (“Big, Medium, Little”) infrastructure, composed of heterogeneous architectures, and a scheduling framework dealing with energy proportionality. We introduce heterogeneous power processors inside datacenters as a way to reduce energy consumption when processing variable workloads. Our framework brings an intelligent utilization of the infrastructure by dynamically executing applications on the architecture that suits their needs, while minimizing energy consumption. Our first validation process focuses on distributed stateless web servers scenario and we analyze the energy savings achieved through energy proportionality. This research activity is performed with the collaboration of Sepia Team (IRIT, Toulouse) through the co-advising of Violaine Villebonnet.

Energy-Aware Server Provisioning

Several approaches to reduce the power consump- tion of datacenters have been described in the literature, most of which aim to improve energy efficiency by trading off performance for reducing power consumption. However, these approaches do not always provide means for administrators and users to specify how they want to explore such trade-offs. This work [27] provides techniques for assigning jobs to distributed resources, exploring energy efficient resource provisioning. We use middleware-level mechanisms to adapt resource allocation according to energy-related events and user-defined rules. A proposed framework enables developers, users and system ad- ministrators to specify and explore energy efficiency and perfor- mance trade-offs without detailed knowledge of the underlying hardware platform. Evaluation of the proposed solution under three scheduling policies shows gains of 25% in energy-efficiency with minimal impact on the overall application performance. We also evaluate reactivity in the adaptive resource provisioning This approach has been applied in the Nuage research project [26] .

Virtual Home Gateway

About 80-90% of the energy in today’s wireline networks is consumed in the access network, , including about 10 to 30W per user being dissipated mostly by the customer premises equipment (CPE). Home gateway is a popular equipment deployed at the end of networks and supporting a set of heterogeneous services (data, phone, television, multimedia, security services). These gateways and associated services can be difficult to deploy and maintain for customers. These gateways are difficult to manage for network operators and consume a lot of energy. We explore the technical solutions to reduce the complexity and energy impact of such equipments by moving services to some external dedicated and shared facilities of network operators. This result is a joint work between Avalon team (J.P. Gelas, L. Lefevre) and Addis Abeba University (M. Tsibie and T. Assefa). This research has been demonstrated in the GreenTouch final celebration event in New York (June 2015).