Skip to Main content Skip to Navigation

Performance evaluation of green IT networks

Abstract : Energy saving in telecommunication networks is a major objective to reduce overall consumption. The IT sector already has a very high contribution to this increase. Indeed, many techniques to reduce consumption in other industries or services results in more IT and telecommunications (the "Green by IT" approach) and therefore in an increase of consumption in IT domains. It is therefore important from an economical point of view to reduce the energy consumption per transmitted or calculated bit (the "Green IT" concept). In the networks domain, energy optimization is mainly based on an adaptation of the architecture and the resources employed according to the traffic flows to be transported and the promised quality of service. We therefore seek to adapt resources to demand, which results in a dynamic dimensioning adapted to the load. This is by nature different from the worst-case dimensioning commonly used. In terms of technology, this requires network equipment to have "sleep", "deep sleep" or "hibernate" modes (terminology varies among suppliers), but all of these modes are associated with the same concept: putting the equipment to sleep to reduce its energy consumption. For the performance/energy trade-off to be relevant, it seems important to use energy consumption formulas obtained from the network resource utilization. The approaches we propose are based on the theory of queuing networks, Markov chain analysis (analytically by proposing new product forms and numerically by suggesting new resolution algorithms) and the theory of stochastic comparison.At the application level we have addressed various issues: DVFS mechanisms with a change of processors speed, task migration between physical servers in a data center (load balancing, consolidation), optical networks with efficient filling of optical containers, intermittent power distribution in a sensor network (and LoRa network) including a new model of Energy Packet Networks (EPNs).
Complete list of metadata
Contributor : Abes Star :  Contact
Submitted on : Monday, May 3, 2021 - 10:48:08 AM
Last modification on : Thursday, November 25, 2021 - 11:04:02 PM
Long-term archiving on: : Wednesday, August 4, 2021 - 6:38:01 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03215137, version 1


Youssef Ait El Mahjoub. Performance evaluation of green IT networks. Networking and Internet Architecture [cs.NI]. Université Paris-Saclay, 2021. English. ⟨NNT : 2021UPASG011⟩. ⟨tel-03215137⟩



Record views


Files downloads