Fig 1 - uploaded by K. K. Ramakrishnan
Content may be subject to copyright.
Source publication
To address reliability challenges due to failures and planned outages, Internet Service Providers (ISPs) typically use two backbone routers (BRs) at each central office. Access routers (ARs) are connected to these BRs in a dual-homed configuration. To provide reliability through node and path diversity, redundant backbone routers and redundant tran...
Contexts in source publication
Context 1
... PoP, again to provide redundancy, and achieve the necessary level of service availability. ARs that are located within the PoP are connected to the two BRs that are in the same PoP. This configuration is also the case for ARs that are close to the PoP. However, ARs that are farther away from a PoP are typically dual-homed to two different PoPs. Fig. 1 shows the configuration within a PoP, where there are two BRs at the PoP that are interconnected using short-range fiber. Further, the two ARs (representative of the multiple that may exist at that PoP) aggregate all the traffic in that metropolitan area, and are dual-homed to these two BRs. The BRs in the backbone at different PoPs ...
Context 2
... first option (Option-1) (see Fig. 10) for reducing the cost of a backbone is to eliminate one BR from each PoP, thus avoiding the cost of the additional BR. While this may be a simple approach, we still need to ensure that this is done in a manner that the availability of the service provided by the backbone network is not adversely impacted. Of the two and in each ...
Context 3
... improve performability, our second option (Option-2) (see Fig. 11) improves on Option-1 by adding a link from each AR to an additional BR, located in a remote PoP (called remote BR, in the rest of the paper). While this does save the cost of the intra-office links, it results in increased transport cost for connecting the ARs to the remote BRs. It also saves the chassis cost of the eliminated s, but ...
Context 4
... final option (Option-3) (see Fig. 11) improves on Option-2 dynamically by setting up an AR-remote BR link upon failure of the local BR. We first eliminate the router from each office, and identify a remote BR for each AR. However, instead of setting up permanent full capacity AR-remote BR links (as in Option-2), we size these links dynamically upon failure of the local ...
Context 5
... propose to use a service platform similar to that utilized by AT&T's GRIPhoN project [6]. A simplified diagram of a PoP is shown in Fig. 12. For simplicity, we show only one AR located in the PoP even though in reality we have several ARs 1 In telecommunications terms, a flap refers to an intermittent failure, characterized by frequent, short duration failure conditions with self ...
Context 6
... additional individual circuits to it as needed. We exploit the OTN layer for sub wavelength circuits (e.g., for setting up the initial low rate 1 G, ODU-0, connection, as in [6]), and the DWDM layer is used for adding wavelength connections, e.g., multiples of 40G. We use an FXC to reuse the ports that are on the AR to the local BR. As shown in Fig. 12, upon failure of the BR at PoP A, all the ports on the access router at PoP A are connected to the BR at the remote PoP ...
Context 7
... and 40G regenerators used along the WDM links of the circuit's path. Each WDM link in the path can either be a Direct WDM link or an Express WDM link (see Section II-B-5). A 40G transponder is used on each end-point of the circuit, and a 40G regenerator is used for interconnecting two adjacent WDM links in the end-to-end path. For example, in Fig. 13, a new 40G circuit is carried over two Express WDM links (curved lines) followed by a Direct WDM link (straight line). Hence the cost for the circuit is . 2) Cost of a 10G Circuit: Computing the cost for a 10G circuit is a bit more involved because a 10G circuit is often carried over a multiplex link (see Section II-B-3). Deploying a ...
Context 8
... Scenario 1 (see Fig. 14), the new 10G circuit uses a new multiplex link carried over two Express links, and a Direct link using an unused wavelength on each link. The wavelength is operated at 40Gbps, and muxponders are used at both ends to carry the new 10G circuit. The total cost for carrying the 10G circuit is . Note that three more 10G circuits can be ...
Context 9
... Scenario 2 (see Fig. 15), the new 10G circuit is carried over an existing multiplex link. The total cost for carrying the 10G circuit is . In Scenario 3 (see Fig. 16), the new 10G circuit is carried over a pre-existing multiplex link, and a new multiplex link that spans two Express links and a Direct link. An unused wavelength at 40Gbps is used on each WDM ...
Context 10
... Scenario 2 (see Fig. 15), the new 10G circuit is carried over an existing multiplex link. The total cost for carrying the 10G circuit is . In Scenario 3 (see Fig. 16), the new 10G circuit is carried over a pre-existing multiplex link, and a new multiplex link that spans two Express links and a Direct link. An unused wavelength at 40Gbps is used on each WDM link, and muxponders are used at both ends to carry the new 10G circuit (similar to Scenario 1). The total cost for carrying the 10G circuit is ...
Context 11
... in Scenario 4 (see Fig. 17), the new 10G circuit is carried over two pre-existing multiplex links. The total cost for carrying the 10G circuit is . Through a combination of the above scenarios, additional options are possible for carrying a new 10G circuit over existing and new multiplex links, and their costs should be calculated ...
Context 12
... We stopped when either this number became zero (all links had utilization between 90% and 110% upon a single failure), or subsequent iterations stopped reducing this number. We also replaced any set of four 10G circuits (in the same logical link) with a 40G circuit. This network design is referred to as the BL, Baseline design, in Section VI. Fig. 18 provides a high level overview of the approach we use to design the initial network topology, as well as our approach to evaluate the proposed ...
Context 13
... the th-remote BR, then we need transport capacity (equal to the amount of priority traffic from -th AR) to set up this link upon failure of the local BR. Transport cost includes the cost of regenerators, fiber, ROADMs, and amplifiers. 4) We may also need additional capacity at certain inter-office BR-BR links. Consider the following scenario (see Fig. 19). Suppose we decide to map the AR at to the remote BR at , and let be a different location. When the local BR at fails, all the traffic that was flowing between and now shifts, and is carried between and . It is possible that some of the links in the to path do not have enough capacity to carry all this traffic, and would require ...
Similar publications
Because of various ecological, environmental, and economic issues, energy efficient networking has been a subject of interest in recent years. In a typical backbone network, all the routers and their ports are always active and consume energy. Average link utilization in internet service providers is about 30-40%. Energy-aware traffic engineering a...
Intradomain traffic engineering (TE) has become an indispensable tool for Internet service providers (ISPs) to optimize network performance and utilize network resources efficiently. Various explicit routing TE methods were recently proposed and have been able to achieve high network performance. However, explicit routing has high complexity and re...
![Fig. 1. DiffServ core router architecture following [2].](profile/Emmanuel-Lochin/publication/322515457/figure/fig1/AS:769220740644864@1560407924751/DiffServ-core-router-architecture-following-2_Q320.jpg)
We define a novel core network router scheduling architecture to carry and isolate time constrained and elastic traffic flows from best-effort traffic. To date, one possible solution has been to implement a core DiffServ network with standard fair queuing and scheduling mechanisms as proposed in the well-known “A Differentiated Services Code Point...
Recent work on Information Centric Networking enables the exploitation of the caching resources of the new generation of routers [Content Routers (CR)]. So far, only a basic Least Recently Used strategy implemented on every CR has been proposed. In this paper, we introduce a cooperative caching strategy in the context of a Content-Centric-Network (...
To address the reliability challenges due to failures and planned outages, Internet Service Providers (ISPs) typically use two backbone routers at each central office to which access routers connected in a dual-homed configuration. At the IP layer, redundant backbone routers and redundant transport equipment to interconnect them are deployed, provi...
Citations
... The WDM layer is composed of regenerators, transponders and OXCs (optical cross connects) and it routes traffic across the WDM links. When the upper IP layer changes the capacities of composite links, the WDM layer might adjust its routing of WDM traffic to different WDM links and change the regenerators' power status (on or off) by the methods in [7] to achieve energy savings. The fiber layer is composed of amplifiers and some underlying elements. ...
We propose a multi-layer energy saving technique for optical core networks that aims at reducing energy consumption by powering off components in different layers of the network After obtaining satisfactory results in saving energy by powering off ports in the IP layer in our previous work in this paper we target more savings by considering additional layers in the network. The model proposed in this paper is a heuristic that bases the capacity prediction for a future time slot on the number of 40G links needed in the current time slot. It also revolves around four parameters for which the values are empirically set. We set two thresholds, low and high, as well as the number of links to power off or power on each time the utilization is below or above a threshold. We assess our model through experiments featuring an Internet2-like topology and a real one-day worth of traffic split into five-minute time slots. The results offer a comparison between different parameters settings and how they affect energy savings and the number of overflows in the network that result from mis-prediction. That said, we demonstrate that our model can achieve up to 90% reduction in energy consumption in the best case when the future traffic is known; otherwise, the savings can range between 82% and 88% with the occurrence of a small amount of traffic overflow events.
Telecommunication networks often have structures that can be leveraged for efficient modeling, yet often are not; and these networks require analyses considering measures of performance beyond what most research enables. In this paper, we take advantage of that structure to show how we can make any path search algorithm more efficient, and apply those results in an extended measure of service and network performance which includes probabilistic measures including reliability, availability, and normalized performability. We further extend to a measure for comparison of networks which relates the correlation of connection points on the network, and show why the measure and comparison is widely useful. The work presented in this paper is easy to understand and apply to a broad range of Information and Communications Technology network design problems and approaches in use.
