Escolar Documentos
Profissional Documentos
Cultura Documentos
AbstractAchieving energy efficiency has recently become an devices enables greater energy savings, in this work we do
essential aim of networking research due to the ever increasing not consider this possibility because of resiliency concerns
power consumption and CO2 emissions generated by large data in case of network events. However, due to the typical
networks. For this problem, the emerging paradigm of Software-
Defined Networks (SDN) can be seen as an attractive solution. In over-provisioning considered in the design and operation of
these networks an energy-aware routing model could be easily backbone networks, considerable energy savings could be
implemented leveraging the control and data plane separation. reached changing the status of network interfaces to sleep
This paper addresses the problem of optimizing the power mode whenever a link is not transferring data.
consumption in SDN using an energy-aware traffic engineering In this context, the optimization of power consumption
approach that minimizes the number of links that can be used
to satisfy a given traffic demand. Different from previous works, could be considered as one promising field of application
we focus on optimizing energy consumption in OpenFlow net- for Software-Defined Networks (SDN). In OpenFlow net-
works with in-band control traffic. Our approach also considers works [6], control functions, decoupled from forwarding de-
performance constraints that are crucial in the correct operation vices, are logically centralized in a new entity called controller.
of SDN, such as bounded delay for the control plane traffic and The logically centralized control plane in SDN has a global
load balance between controllers. First, we present a complete
formulation of the optimization problem involving the routing knowledge of network state information. Furthermore, it can
requirements for control and data plane communications. To manage network tasks and perform device configuration with-
reduce the time complexity of our model in large-scale topologies out the need of additional software or hardware in each one of
we derive a heuristic algorithm. Significant values of energy the switching elements. Meanwhile, the network devices only
saving (up to 60%) are reached in the simulations using real forward traffic according to the rules set by the controller. This
topologies and demands data.
feature can be leveraged to perform an energy-aware route
scheduling that determines, in a coordinated and centralized
I. I NTRODUCTION
way, the switch interfaces that should be put to sleep. There-
Energy consumption of Information and Communication fore, an energy-aware solution could be easily implemented
Technologies (ICT) has become a concerning issue in the in the control plane. In this paper we address the problem of
last decade. According to [1] in 2012, close to 4.7% of the optimizing the power consumption in SDN by minimizing the
worlds electrical energy was consumed by ICT, releasing into number of active links needed to route a given traffic demand.
the atmosphere roughly 1.7% of the total CO2 emissions. Different from previous works which focus on power-
Moreover, recent studies state that energy demand of ICT minimization in SDN considering only the data plane traffic,
sector is growing faster than the overall one [2]. Among the we aim to optimize energy consumption in OpenFlow net-
main ICT sectors, telecommunication networks account for works with in-band control traffic. In this way, the energy-
more than a third part of the ICT total energy consumption [3]. aware routing performance can be analyzed when, for phys-
As a result, the reduction of power consumption in Internet ical and cost restrictions, implementing a dedicated control
Service Provider (ISP) backbone networks is crucial to accom- network is not feasible. Furthermore, this is a more realistic
plish significant energy savings in this sector. At the same scenario for large backbone networks, where dedicated links
time, increasing the energy efficiency in data networks can to transfer the control messages between controllers and
also substantially reduces the environmental impacts of other forwarding devices are impractical and cost-inefficient.
sectors. Our approach also considers performance constraints that
Given that the energy consumption of network equipment are crucial in the correct operation of SDN, such as bounded
is only slightly influenced by their traffic load, an effective delay for the control plane traffic and load balance between
strategy to minimize the consumption of data networks is to controllers. Although this is a more challenging approach, our
reduce the number of active elements [4]. This feature can be model enables the reduction of power consumption without
implemented by putting into a low-power sleep state (sleep performance degradation.
mode) elements such as line cards or port interfaces that are The energy-aware approach proposed in this work is more
not in use [5]. Although turning off entire interconnection fine-grained since it includes integrated routing considerations
20 1 0.6351166 0.02545899
2 0.5631809 0.03049248
Abilene
3 0.4893286 0.03668678
10 4 0.4268516 0.04307767
1 8.068126 0.1452133
2 6.307062 0.1227532
Nobelus 3 4.995122 0.1465722
1 2 3 4
4 4.150009 0.1727835
Number of Controllers
(a) Average of shutdown links in the Abilene topology. 5 3.445598 0.2240748
1 3446.204 305.5014
Geant
Optimal
2 2616.54 53.60402
40 Heuristic
Average Energy Savings (%)
50
40
40
30
30
20
1 2 3 4 5 0 2 4 6 8 10 12 14 16 18 20 22 24
Number of Controllers Daily Hour
Fig. 3. Average energy savings in the Geant topology varying the number of Fig. 4. Average energy savings in the Geant topology with two controllers
controllers. during a day period varying the control paths delay bound.
processing time increases dramatically. For instance, in this denote the relation among and the shortest path propagation
topology the ILP model can take more than 3000 s to find latency for every control pair of nodes.
solution while it is almost less than 300 s for the heuristic Fig. 4 shows the average energy savings using the network
algorithm, i.e one order of magnitude improvement. For the diameter () as delay bound against three more restrictive
Geant topology we only show the case of having one and two possibilities ( = 1, 1.5 and 2). For instance, = 2 means
controllers, because beyond this limit the convergence time of that every control path latency is, at most, twice that of the
solving the exact model considering all possible combinations shortest path. Since the traffic offered for Geant topology
of nodes as controllers placement, became unfeasible. is almost constant during a day period, the energy savings
As it is also shown, in both approaches the computation time outlined present very few variations. As expected, less energy
decreases while the number of controllers grows. This is due to is saved when we use only shortest paths (i.e. = 1) to route
the fact that our energy-aware approach avoids the routing of the control messages, but even then, energy savings of 30%
additional traffic load through the controllers. Therefore, with could be achieved. This result shows that our approach enables
the increase of network controllers a fewer number of alternate considerable power savings without degrading the delay of
paths between each pair of nodes need to be considered in the control plane communications.
simulation.
1.0
Energy Savings
35
Control Path Delay mode. In this way, an energy-aware control plane could be
achieved. As future work, we plan to extend this work to take
into account the use of restoration mechanisms in order to
0.8
the comparison of this model with existing solutions will be
Average Energy Saving (%)
30
0.6
ma y Competitividad of the Spanish Government under
25
0.4
20