Telecommunication Network Engineering Fast ...

Telecommunication Network Engineering

Fast Reroute technique for Energy Efficiency Diego Reforgiato Recupero University of Catania (Italy), [email protected] Sergio Consoli Brunel University (United Kingdom), [email protected]

Consideration regarding the use of the Fast Reroute technique for Energy Efficiency

Abstract This article is addressed to network engineers and managers of large enterprises that are interested in energy consumption of network devices. In the last years many efforts in telecommunications network research and design have been devoted to reduce energy consumption of network devices. Development of green routers aimed at saving energy when the input traffic is low is receiving increasing attention in the scientific community. In this paper we face the problem of further reducing power consumption in transport networks adopting the Fast Reroute technique together with a low power idle approach, which result in substantial energy saving when the traffic on the network is low. We have run simulations using NS-2 and as test case, we developed the proposed approach on top of the NetFPGA architecture and results are very promising for larger exploitation. Keywords: Fast Reroute · Energy Saving · MPLS · NetFPGA - Router


Introduction Today, the load of networks typically exceeds their long-term utilization by a wide margin (Reforgiato, 2013). Moreover, as shown in (Barford, 2010), current network nodes have a power consumption that does not depend on the actual traffic load they face. This implies an energy waste in today networks (Ipmon Sprint, 2007) (Jardosh, 2007). The increasingly cost of energy led to a general concern about this occurrence. Today, in fact, 37% of the total ICT emissions are due to telecommunications companies infrastructures and devices. For this reason, addressing energy efficiency challenges in wireline networks is receiving considerable attention in the literature (Barford, 2008) (Gupta, 2003) (Mahadevan, 2009) (Nedevschi, 2008). Moreover a number of research projects have been started on this topic (see for example (Earth, 2010), (Chron, 2010), (Greentouch, 2011), (C2power, 2010)). Some novel hardware devices that allow entering different energy power states are therefore expected in the near future (Cisco, 2009). Routers are device that forwards data packets between computer networks, creating an overlay internetwork. Cisco (Cisco, 2009), the pioneer in home networking, has sold more than 70 million Linksys routers worldwide and is leading the charge in the next-generation connected home. A routing protocol specifies how routers communicate with each other, disseminating information that enables them to select routes between any two nodes on a computer network. Routers are therefore one of the core elements of computer networks. Moreover, Multiprotocol Label Switching (MPLS) is a mechanism in high-performance telecommunications networks that directs data from one network node to the next based on short path labels rather than long network addresses, avoiding complex lookups in a routing table.


Our idea builds upon several approaches that have been proposed to reduce the energy consumption in network components. For example, in (Nedevschi, 2008), the authors proposed a method based on putting network components to sleep during idle times, reducing energy consumed in the absence of packets. The rationale behind that was based on adapting the rate of network operations to the offered workload, reducing the energy consumed when actively processing packets (rate adaptation is usually achieved by scaling the processing power according to the data rate the router has to manage). Which energy aware technique to use in a green router (a router with novel hardware and software capabilities able to dramatically reduce its power consumption) is a challenging problem which depends on a number of parameters, including the role of the router in the network, the incoming traffic profile, the hardware complexity and the related costs with respect to the energy that can be potentially saved and the Quality of Service (QoS) to be guaranteed to the final users (Hu, 2011). In (Bianzino, 2011), a novel distributed algorithm puts into sleep mode links in an IP-based network. This solution is distributed among the nodes under the assumption that nodes do not know the traffic matrix, whose knowledge is indeed unrealistic in the current Internet architecture. Moreover, the switch off decision is taken considering the current load of links and the history of past decisions. One of the most common policies in network dimensioning is represented by resource consolidation: its goal is to reduce energy consumption due to devices underutilized at a given time. As the traffic in a given network follows a well known daily and weekly pattern, there is an opportunity to aggregate traffic flows over a subset of the network devices and links, allowing other devices to be temporarily switched off. This solution should preserve connectivity and QoS, for instance by limiting the maximum utilization over any link. On this direction, authors in (Bianzino 2011), (Kakemizu, 2009) proposed approaches that achieve energy efficiency in networks by


controlling the usage probability of pre-defined multiple paths to bring about a situation where some nodes can transition to sleep state when the traffic volume is small. Moreover, stand-by modes have to be explicitly supported with special techniques to maintain the network presence of sleeping components. Last but not least, in (Bianzino, 2010) the authors analyze the design of green routing algorithms and evaluate the achievable energy savings that such mechanisms could allow in several realistic network scenarios. In this chapter we face the problem of reducing power consumption in router networks adopting a Low Power Idle1 (LPI) approach locally for each router. By sending a low-power-idle indication signal for a specified time the transmit chips in the system can be turned off. We exploit the Fast Reroute mechanism, which is a technique, integrated in the MPLS protocol, that detects link failures at the hardware level (without Fast Reroute the OSPF protocol would require several seconds) and overwrites respective routing table entries with pre-computed routes, to minimize packet drops. An initial idea has been presented in (Reforgiato, 2013b). We focus on wired networks and assume that interfaces have a certain number of LPI states: each of them is characterized by a maximum bit rate that can be supported and a given power consumption. The goal is to maximize the number of interfaces working in LPI mode (which corresponds to a certain amount of energy saving) in the network routers maintaining the reachability of all potential destinations and maintaining high the network QoS. More in detail, we have leveraged the Fast Reroute approach in order to detect the hardware interfaces, which can be set to LPI mode (therefore obtaining energy saving). The output traffic of the underlying router can be therefore deviated (depending on the current bit rate and a local control policy)

1

http://en.wikipedia.org/wiki/Energy-‐Efficient_Ethernet


towards other interfaces producing a non-negligible gain of energy saving. We have run simulations using NS-2 (a discrete event simulator targeted at networking research) to simulate the behavior of the nodes that use our approach within several network topologies. The results we have obtained showed us the amount of energy saving that is possible to achieve using our method. The results drove us to the development of such a technique on a real environment. In fact, we built our technique on top of a project based on the reference router of the NetFPGA platform (Gibb, 2008). The remainder of this paper is organized as follows: in the next Section we described how we employed the Fast Reroute mechanism for energy efficiency. Then we will show how to detect the interfaces to be set to LPI mode. A result section will show experimentations, simulations with NS-2, and results. It follows an analysis on the NetFPGA platform, which is the case study we have been working on. Measurements relative to power consumption of the NetFPGA platform are reported, where we have estimated the amount of energy that could be saved using our approach and the NetFPGA power consumption analyzed in (Lombardo, 2012). Finally we will show the future research direction before ending the paper with conclusions and key terms definitions.

2 Background In a networking environment, a router must be able to forward packets to each destination. A common router keeps active all its interfaces in order to find the best path for each destination and minimize the cost of packet transmissions. Network operators tend to keep this cost as low as possible. Our goal is to maximize the number of interfaces set in LPI mode in the network routers maintaining the reachability of all potential destinations and maintaining high the network QoS. To know which interfaces can be set to LPI mode we need to understand how the


network topology is characterized and how each router can forward packets to all destinations. Some of the information about the topology can be obtained by looking at the info provided by the routing protocol. For example, in the link state routing protocols, a router gathers information about the entire network topology, and the Djikstra’s algorithm (Diestel, 2000) is performed using this information in order to calculate the best path for each destination. Moreover, if a given destination can be reached via multiple paths with the same cost, load balancing can be applied to improve the QoS of the communications. The reader notices that not all routing protocols implement the load balancing capability. Using this process the router can detect its output links where to forward a packet to reach a desired destination. We propose a solution to detect which interfaces can be set to LPI mode, allowing us to reduce the energy consumption and, at the same time, keeping high the QoS. We have started our work with the analysis of the information about the topology, which is collected by the OSPF routing protocol and the Fast Reroute technique. The Fast Reroute technique routes packets over alternative paths to minimize packet drops. These alternative routes are pre-computed and previously included in the routing table by the router software. With such a mechanism, each router is aware of at least two different routes, if they exist, (which use two different interfaces) for each destination. Using the fast reroute technique it is possible to detect multiple paths to reach each node of the network. Using this kind of information, our approach is able to aggregate out-going traffic and forward it towards a small set of interfaces working in a default behavior, whereas the other interfaces of the router can be set to LPI mode. A Network Control Unit (NCU) will manage the power state of these.


Fig. 1 – Example topology

3 The core of our approach As previously stated, in order to identify the interfaces to be set to LPI mode, we need to detect the indispensable interfaces, that is, the ones that need to stay in the normal operational state (no LPI mode) in order to work with the highest bit rate that the interfaces hardware supports guaranteeing the network presence of the considered router. Let us consider the example topology shown in Figure 1. This network contains 10 nodes. We will focus our analysis on the node X, which can send anything to nodes 1-4. X has four interfaces (Eth1, Eth2, Eth3, Eth4); they connect X directly to the nodes 1, 2, 3 and 4, respectively. In a scenario where a single-path routing protocol is used, each router saves the routing information in tables similar to Table 1 where, for each destination d, it is known the interface (where the corresponding entry is equal to 1) to be used to forward traffic to d.


Table 1 – Forwarding table of node X with a single-path routing protocol.

Table 2 – Forwarding table of node X using the fast reroute approach. Basically, each row in the table corresponds to a destination. Each entry of the table contains: – 1 If the corresponding destination is directly reachable through the interface specified in the corresponding column; – 0 otherwise. Thus, the routing protocol is aware of a single interface to reach each destination of X. This is the huge limit for single-path routing protocols that do not allow inferring anything about the other interfaces nor applying the green energy efficiency technique we propose. When the Fast Reroute approach is used, for the node X we obtain the routing information shown in Table 2. With such a protocol, each node knows up to two different paths to reach any other destination: the main path and the backup path. The goal is to find a routing table similar to Table 3.


Table 3 – Output forwarding table.

3.1 Network Control Unit Each router is equipped with a Network Control Unit (NCU) which checks, for all the nonindispensable interfaces, whether they cannot sustain the overall incoming bit rate of the aggregated traffic. For such interfaces, if the traffic is above (below) a certain threshold that a given interface in the current LPI mode can support, then the NCU sets the interface to the next (previous) LPI state. The correspondent routing entries are thus overwritten in order to use the alternative path to reach the destination. These entries will be restored to the initial value by the NCU itself according to the traffic load or the routing updates information from the network. Using this approach, the non-indispensable interfaces are used only to receive network packets and the network presence is guaranteed.

4 Experimentation In this section we will discuss about the energy efficiency improvement obtained using the technique we have proposed and tested using network simulator NS-2. As each different router has a different power consumption depending on different hardware components, for our analysis we will consider the power consumption measurements discussed for the NetFPGA in


Section 5.2. We run simulations with the primary goal to estimate the average energy gain obtained when setting the interfaces to LPI mode. Following this direction, we have initially simulated a random generated topology. The traffic has been sent through this synthetic network. We have thus compared the power gain obtained using our method with respect to not applying any energy saving techniques.

Fig. 2 – Resulting topology with two interfaces (Eth2 and Eth4) set to LPI mode

Moreover, we have analyzed the behavior of the network varying the amount of traffic that each node has to forward to the destinations. With all this in mind, we have analyzed the topology composed by two kinds of nodes: – the core nodes, that constitute the trunk of the network; – the external nodes, which are the sources and the destinations of traffic flows forwarded through the trunk network. For each simulation run, the created topology is composed by 30 core nodes and 30 external nodes: the core nodes are randomly connected to each other, while the external nodes are connected to exactly one core node, randomly chosen. Therefore, for each run, we obtained a new strongly connected topology. The NCU checks whether those interfaces set to a given LPI


state can sustain the incoming traffic. If not, it switches them to next available LPI state until they are switched back to the default behavior. The core nodes are connected by full duplex links of 1 Gbps. On the average, there are 62 links in the interconnection network of the core nodes. Therefore the dimension of the network we are considered is of 60 nodes with approximately 62 links. As stated above, each external node is connected to exactly one core node and, through such node, it sends a traffic flow to another random external node. Each external node is, at the same time, the sender of a traffic flow and the receiver of another traffic flow. In order to evaluate the bit rate that is possible to support with our technique without incurring in QoS degradation, we have run our simulation several times, therefore obtaining results for different topologies. In particular, each simulation is divided in slots of 20 seconds. During each time slot source nodes send data at constant bit rates. Moreover, each time slot is characterized by a different transmission bit rate for each external node (from 100 Mbps to 1 Gbps, with a step of 100 Mbps). Therefore, during the first time slot all source nodes send data at 100 Mbps, during the second time slot sources nodes send data at 200 Mbps, and so on. In terms of power gain, on the average and with the generated topologies, the current results show that about 46% of the interfaces have been marked as non-indispensable (and therefore set to LPI mode) using our technique and, consequently the power consumption has been decreased according to the local control policy of each NCU which according to the current bit rate chooses the LPI state of such interfaces. We have compared the energy saving when all the interfaces are set to the default behavior (no LPI mode used) and when all the non-indispensable interfaces are set to LPI mode. Figure 3 shows the average power consumption in Watt for all the interfaces in our topology. It is Watt if all the interfaces work at their default behavior. Using our approach, we are able to set to LPI mode about 40 interfaces of the network.


Fig. 3 – Power consumption of the network interfaces. The reader can notice that using our approach always results in a power gain. The lower the current bit rate of each node, the higher the power gain we can achieve. A scenario where the overall bit rate is lower than the 30% of the entire network capacity is similar to real networks in the nighttime where the application of our technique could save a great amount of energy. The histogram in Figure 4 shows the average number of interfaces set to the default behavior for the network topology used in our simulations. The blue bar indicates the average number of interfaces set to the default behavior (no LPI mode) when our approach is not used. Conversely, the red bar indicates the average number of interfaces set to the default behavior using our approach.


5 Case study 5.1 NetFPGA implementation of the proposed method The NetFPGA (NetFPGA, 2007) card is embedded within a host computer, where it is executed PW-OSPF, a user level router that performs IPv4 forwarding, handles ARP and various ICMP messages, has telnet (port 23) and web (port 8080) interfaces to handle router control, and also implements a subset of OSPF named PW-OSPF. Moreover, the routing protocol mirrors a copy of its MAC addresses, IP addresses, routing table, and ARP table to the NetFPGA card which hardware accelerates the forwarding path. PW-OSPF is a greatly simplified link state routing protocol based on OSPFv2. Like OSPFv2, routers participating in a topology have to periodically broadcast HELLO packets for the discovery and the connection management of the neighbors. Whenever any change occurs in the topology, like the addition or deletion of a router, or when a timeout occurs, each router floods its routing table information in order to provide to each other router a complete vision about network topology. According to this information each router performs Djikstra’s algorithm independently to determine the next hop in the forwarding table, obtaining the best path. PW-OSPF creates routing table providing only one entry for each destination, because it does not implement load balancing. According to what explained in Section 3, we need to couple the classical OSPF with the Fast Reroute technique. We based our work on the raw implementation of the Fast Reroute technique, which exists within one of the NetFPGA projects developed at Stanford University.


Fig. 4 – Average number of interfaces in the network set to the default behavior. The Fast Reroute and Multipath Router (FRMP) project is built on top of the reference router with major changes to the output port lookup module. The project aims at implementing two important features of MPLS protocol on the NetFPGA architecture: Fast Reroute and Multipath Routing. Moreover, an improved routing protocol, which implements the Fast Reroute and the Multipath Router, is run. Fast Reroute (Pan, 2005) detects hardware link failures or topology changes. The reference router is generally based on the OSPF messages timing out. However, this causes packets to be dropped in the interval between the actual failure and failure detection. From this point of view, Fast Reroute detects link failures at the hardware level and forwards packets over alternative routes to minimize packet drops. These alternative routes are precomputed by the router software and used immediately, if necessary. We have developed on top of the Fast Reroute and Multipath Router project a software module in ANSI C, which connects to the process, which handles the routing protocol and enables the Fast Reroute feature. Such software receives all the routing table information and builds a forwarding table similar to the ones previously used in Section 3. As in the NetFPGA platform it


is not possible to set the LPI mode for any of the four 1G-Ethernet ports, we considered four LPI states for each interfaces corresponding to four bit rate levels: 250Mbps, 500Mbps, 750Mbps, 1Gbps. The first state allows the interface to support up to 250Mbps consuming 25% of the power consumption when working in the default behavior. The second state allows the interface to support up to 500Mbps consuming 50% of the power consumption and so on. Our software periodically issues the commands to set to LPI mode the interfaces that will not be used to send traffic; then the NCU manages the considered LPI mode of each interface choosing which of the four state the interface should be on, and, if needed, sets them back to the default behavior. In order to have a real working prototype, it is needed to provide the NetFPGA Ethernet devices with LPI mechanism. In our analysis we have simulated the desired behavior.

5.2 Power Consumption in NetFPGA platform In this section we will consider the power consumption of the NetFPGA platform, which explains some of the results we showed in Section 4. Table 3 reports the power consumption of the NetFPGA measured in (Lombardo, 2012). The NetFPGA platform may work at either 125MHz or 62.5MHz according to a hardware register CPCI_CNET_CLK_SEL_REG and the local control policy implemented (see (Lombardo, 2012b) for further details);

Table 3 – Measures taken from (Lombardo, 2012) of the power consumption for different number of active ports for the two different values of f


The histogram in Figure 5 shows the average power consumption of the network topology used in our simulations when our approach is not used (blue bar) or it is (red bar). The reader notices that to come up with such values we have considered the power consumption relative to the NetFPGA when on the average 20 routers work at 125 MHz and the others work at 62.5 MHz. With these settings the average energy consumption for the network was about 420 W (blue column). Then, we have subtracted from the result the amount of 80W and show that in the red column, that is the energy saving relative to the 40 interfaces, which have been set to LPI mode using our technique.

Fig. 5 – Average power consumption of the network topology used in our simulation 6 Future Research Directions The increase of CO2 emissions triggered concerns related to the public health. This led to the start of the green industrial revolution. Besides, the increase of the energy costs has brought further interest in techniques for energy saving worldwide. The ICT sector is moving fast in order to adopt green solutions in every domain, including the telecommunication networks. This chapter has described one energy efficiency technique for wired networks, which employs a new


software approach on top of the fast reroute protocol. New programmable routers, like the NetFPGA 10G, have been released in order to easily test new methods and applications. Before the development of such devices, within the telecommunication networks domain, one had to simulate new protocols and approaches he would come up. Then, only based on simulation results, big telecommunication industry developed new hardware. Today, with the widespread and use of programmable routers, it is possible to provide anyone with a device that is possible to program and test.

7 Conclusions The target of this paper is the definition of a new active sleeping technique in wired networks, which sets the interfaces of a router to a LPI mode and forwards the traffic towards the indispensable interfaces only. Our technique uses network-wide information about the current topology and then it acts locally in each router without affecting the operational behavior of the other routers (which may not use the proposed technique). This solves potential compatibility issues that may arise. Moreover, a Network Control Unit checks for all the non-indispensable interfaces, whether the current interfaces cannot sustain the overall bit rate. In that case, if the traffic is above (below) a certain threshold that a given interface in the current LPI mode can support, then the NCU sets the interface to the next (previous) LPI state avoiding potential loss. We have taken the NetFPGA platform as a case study, and conducted an extensive measurement campaign on the NetFPGA open router in order to analyze the power consumption using our technique. We have simulated our technique onto a realistic network scenario and have shown the achievable power gain we are able to get. In the future we will analyze wireless protocols in order to adapt our technique in wireless environment. In addition, the system is fully


generalizable and in the future it is expected to extend its functionality. Employing further routing techniques, appropriate handling of nested collocations, reducing processing time, and conducting user-centric tests, are interesting topics currently under study.


References Barford, P., J. Chabarek, C. Estan, J. Sommers, D. Tsiang, and S. Wright. (2008). Power awareness in network design and routing. In Proc. of IEEE INFOCOM, Phoenix, USA, 2008.

Bianzino A.P., C. Chaudet, F. Larroca, D. Rossi, and J. P. Rougier. (2010). Energy-aware routing: A reality check. In Proc. of GLOBECOM Workshops (GC Wkshps), IEEE, pages 1422–1427.

Bianzino A.P., L. Chiaraviglio, and M. Mellia. Grida. (2011). A green distributed algorithm for backbone networks. In Online Conference on Green Communications (GreenCom), 2011 IEEE, pages 113 –119.

C2power. (2010). http://www.ict-c2power.eu/.

Chengchen Hu, Chunming Wu, Wei Xiong, Binqiang Wang, Jiangxing Wu, and Ming Jiang. (2011). On the design of green reconfigurable router toward energy efficient internet. In Communications Magazine, IEEE, volume 49, pages 83–87.

Chron. (2010). http://www.ict-chron.eu/content/partners.aspx. 2010.

Cisco. (2009). Ciscoenergywise,http://www.cisco.com/. 2009.


Diestel R. (2000). Graph theory. Springer-Verlag, New York.

Earth. (2011). https://www.ict-earth.eu/default.html. 2011.

Gibb G., J. W. Lockwood, J. Naous, P. Hartke, and N. McKeown. (2008). Netfpga: An open platform for teaching how to build gigabit-rate network switches and routers. In IEEE Transactions on Education, August.

Greentouch. (2011). http://www.greentouch.org/.

Ipmon Sprint. The applied research group. http://ipmon.sprint.com/. 2007.

Gupta M. and Suresh Singh. Greening of the internet. (2003). In Proceedings of the conference on Applications, technologies, architectures, and protocols for computer communications, SIGCOMM ’03, pages 19–26, New York, NY, USA. ACM.

Jardosh Amit P., Gianluca Iannaccone, Konstantina Papagiannaki, and Bapi Vinnakota. Towards an energy-star wlan infrastructure. (2007). In Proceedings of the Eighth IEEE Workshop on Mobile Computing Systems and Applications, HOTMOBILE’07, pages 85–90, Washington, DC, USA.

Kakemizu M. and A. Chugo. Approaches to green networks. Fujitsu Sci. and Tech. J. (2009). 45(4):298–403.


Lombardo A., Carla Panarello, Diego Reforgiato, and Giovanni Schembra. (2012). Measuring and modeling energy consumption to design a green netfpga giga-router. In Proceedings of IEEE Globecom 2012 Conference.

Lombardo A., Carla Panarello, Diego Reforgiato, and Giovanni Schembra. (2012b). Power control and management in the netfpga gigabit router. In Future Network & Mobile Summit (FutureNetw), 2012, pages 1–8.

Mahadevan P., Puneet Sharma, Sujata Banerjee, and Parthasarathy Ranganathan. (2009). A power benchmarking framework for network devices. In Proceedings of the 8th International IFIP-TC 6 Networking Conference, NETWORKING ’09, pages 795–808, Berlin, Heidelberg.

Nedevschi Sergiu, Lucian Popa, Gianluca Iannaccone, Sylvia Ratnasamy, and David Wetherall. (2008). Reducing network energy consumption via sleeping and rate-adaptation. In Proceedings of the 5th USENIX Symposium on Networked Systems Design and Implementation, NSDI’08, pages 323–336, Berkeley, CA, USA.

NetFPGA (2007). Netfpga team. netfpga website. http://netfpga.org.

Pan P., G. Swallow, and A. Atlas. (2005). Fast reroute extensions to rsvp-te for lsp tunnels. RFC 4090 (Proposed Standard), Internet Engineering Task Force, May.


Reforgiato. D. (2013). Toward a Green Internet. Science Mag. 29 March 2013. Vol. 339, no. 6127, pp. 1533-1534.

Reforgiato. D., V. Riccobene. (2013b). FREE – Fast Reroute for Energy Efficiency, IEEE INFOCOM 2013, Poster Session.

ADDITIONAL READING SECTION Fast Reroute and MultiPath Router. https://github.com/NetFPGA/netfpga/wiki/FastRerouteAndMultipathRouter

NS2. http://nsnam.isi.edu/nsnam/index.php/main page.

Ospf version 2. (1998). RFC 2328 (Standards Track), Internet Engineering Task Force, April. http://www.ietf.org/rfc/rfc2328.txt

Pinheiro Eduardo, Ricardo Bianchini, Enrique V. Carrera, and Taliver Heath. (2001). Load balancing and unbalancing for power and performance in cluster-based systems.

PW-OSPF. (2011). http://yuba.stanford.edu/ cs344/admin/pwospf/. Reference router project. https://github.com/NetFPGA/netfpga/wiki/ReferenceRouterWalkthrough


SCONE. https://github.com/NetFPGA/netfpga/wiki/SCONEWalkthrough

Thaler. D. and C. Hopps. (2000). Multipath issues in unicast and multicast next-hop selection. RFC 2991 (Informational), Internet Engineering Task Force. http://www.ietf.org/rfc/rfc2991.txt.

Yamada A., S. Imai, and M. Kakemizu. (2010). A path-based traffic control method for green networks. In Proc. of Information and Telecommunication Technologies (APSITT), 8th AsiaPacific Symposium, pp.1–5.

KEY TERMS & DEFINITIONS Router: A router is a device that forwards data divided in packets among different networks. Router Protocol: A routing protocol specifies how routers communicate with each other, disseminating information that enables them to select routes between any two nodes on a computer network. Quality of Service: The Quality of Service (QoS) is the overall performance of a computer network. Several related aspects of the network service are error rates, bandwidth, throughput, transmission delay, availability, jitter. Green Router: A green router has software/hardware capabilities that reduce its power consumption while maintaining the same QoS. Low Power Idle: LPI is a signal that puts to sleep the transmit chips in a system. MPLS: Multiprotocol Label Switching is a mechanism in high-performance telecommunications networks that directs data from one network node to the next based on short path labels rather than long network addresses, avoiding complex lookups in a routing table.


NetFPGA: The NetFPGA is an open source hardware and software for rapid prototyping of computer network devices.