ECONET will investigate innovative criteria and policies that enable next generation networks and
related equipment to locally and jointly control, manage and monitor energy consumption according
to traffic volumes and performance requirements. This will be achieved by exploiting the novel
green capabilities and technologies resulted from the previous research axis.
Here, the output will be a new integrated framework for flexible and cognitive network management
and operation. Such framework will enable dynamic, ad-hoc and optimized resource allocation,
control, deployment, and administration, at both levels of the individual devices and of the
network.
To this end, as illustrated in Fig. 1, the ECONET project will focus its efforts on:
The development of local control policies (i.e., governors) able to optimally configure and
synchronize the energy-aware capabilities of all the modules and elements in the same network
device or in the same cabinet/Point of Presence;
The introduction and development of novel algorithms and optimization criteria for
network-wide control, which exploit and take green metrics and device capabilities into account;
The design of novel green extensions for network-wide control protocols (i.e., for traffic
engineering and OAM purposes), which will allow to suitably coordinate the energy savings and
performance profiles of each network node;
The conception of new paradigms to continuously monitor network energy consumptions.
Fig 1: Energy-aware control and monitoring strategies proposed by the ECONET project.
Local control policies will be devoted to coordinate the energy profiles and resource allocations
of all energy-aware elements in a device, thus optimising the trade-off between its overall power
consumption and traffic loads/requirements. As shown in Fig. 2, this will be accomplished by
jointly synchronizing and effectively tuning the capacities and the behaviour of processing engines,
network interfaces, etc., which generally constitute the data-plane chain of network devices.
Extensions of local control policies will consider how to jointly reduce the energy consumption of
paired devices (for redundancy purposes) as well as of devices in the same cabinet/point of presence.
Moreover, in some deployments it is essential to be able to limit the overall power consumption of a
system or a set of systems. This can be a result of shortage (temporary or permanent) of energy,
cooling, etc.
Part of the ECONET project is to address this concern and to provide the right abstractions and
interfaces to tune the overall power consumption at the system and network level, to employ the right
policies for such power saving and to optimize the overall power consumption and
functionality/performance under the overall power constraints. Example policies can be changing the
redundancy schemes, reducing the backbone bandwidth, updating routes, etc.
Fig 2: Example of a local control in a crossbar- and linecard-based network device architecture.
Local optimization policies are also expected to provide detailed information and estimates regarding
devices’ heat dissipation. These data will be used in order to finely modulate cooling systems and
thus to further reduce energy consumption (since power and heat management account for ca. 35% of the
total).
Regarding the algorithms and protocols for distributed network control, it is worth noting that telcos
and ISPs already widely adopt frameworks that allow the short- and long-terms monitor and control, but
only on the network performance basis.
Current network-wide control plane frameworks generally use different criteria to allocate network
resources ranging from QoS requirements to bandwidth optimization. In more detail, such frameworks
commonly allow for modifying the way by which traffic flows cross the network (e.g., by changing their
internal paths or by assigning a different QoS level), and can be usually divided in two main
categories: traffic engineering and routing protocols, and centralized OAM.
Traffic engineering and routing protocols (like OSPF-TE) allow autonomic and distributed network
reconfigurations to optimally meet short-term traffic volumes and service requirements. They are
generally needed to better follow the network dynamicity.
In few simple words, centralized OAM systems give to operators the capabilities of:
Monitoring the behaviour of the deployed networks and nodes;
Applying network planning/provisioning algorithms for optimizing the long-term behaviour;
Remotely and statically setting new configurations to network nodes.
In this respect, such OAM systems are usually based on mechanisms and protocols for collecting performance
indexes from network nodes to Network Operations Centers (NOCs), where operators can centrally monitor the
overall network behaviour, and eventually apply user-assisted optimization procedures to every network
device.
Given the differences in their working features and goals, these two network-wide framework typologies are
obviously not exclusive. On the contrary, their joint deployment is needed for an effective and optimal use
of network infrastructures of Telcos and ISPs.
In this respect, the main ECONET objective consists of extending current state-of-the-art technologies,
methodologies and protocols for network-wide control (both distributed autonomic and centralized ones) by
introducing novel energy-aware optimization criteria.
Fig 3: Example of network energy-aware configuration.
As shown in the simple example of Fig. 3, the base concept pursued by ECONET consists of acting at the
network-wide control-plane by changing the paths of traffic flows in order to find the optimal network
configuration, where end-to-end service requirements are satisfied and the overall network energy
consumption minimized.
To this end, beyond classical data on traffic volumes and requirements, such novel mechanisms will have to
know the available green capabilities (i.e., power scaling, smart standby, etc.) and relative features
(e.g., power consumption, etc.) at each node in order to calculate the optimal network configuration. In
more detail, the novel green criteria and mechanism studied by ECONET will aim at providing:
Novel traffic engineering/routing policies to dynamically rerouting traffic for optimizing at
short-term the ratio between performance and energy consumption;
Novel OAM tools for energy-aware long-term configurations of networks and devices;
Novel green OAM criteria for the long-term network design, planning and provisioning, which will take
economic aspects (OPEX and CAPEX) into account.
To support these novel network-wide control criteria, the project will thus develop new energy-efficient
protocols for the dynamic network optimization, which will be devoted to obtain key information on traffic
profiling, power measurements, traffic engineering issues and impact of networking devices on energy costs.
Fig 4: Main goals in the minimization of network energy consumption.
In the same framework, equipment manufacturers will deliver guidelines on how to integrate new device
functionalities in support of the new energy-efficient protocols and hardware elements able to quickly
enter / exit the idle state.
Regarding network-wide monitoring frameworks, they are a crucial part of centralized OAM frameworks, and
are needed for providing input data to the novel green criteria for network design and long-term
configurations.
In this respect, the ECONET project will develop novel network distributed tools able to effectively
collect data on energy consumptions at the device and device component level.
Finally, Fig. 4 tries to summarize the main goals, the different roles and the relationships among
the three energy-aware control-plane typologies proposed by the ECONET project, namely local optimization,
routing and traffic engineering and network-wide monitoring, respectively.
