Conference Paper

A backward-compatible inter-domain multipath routing framework

Tech. Univ. Carolo-Wilhelmina zu Braunschweig, Germany
DOI: 10.1109/INFCOMW.2011.5928794 Conference: Computer Communications Workshops (INFOCOM WKSHPS), 2011 IEEE Conference on
Source: IEEE Xplore

ABSTRACT We present a framework to facilitate the inter-domain multipath routing for carrier networks that are based on circuit switching technologies, without significant changes to the existing inter-domain protocols. We first introduce a simple mechanism, namely, Single Routing Plane Multipath (SRPM) which can be easily implemented in the existing PCE-based inter-domain architectures. The main idea behind the SRPM method is to represent multiple link disjoint paths as a single virtual edge in the PCE and thus use the existing single path routing mechanisms to enable multipath routing. We further present a more generic mechanism, referred to as Multiple Routing Plane Multipath (MRPM) method, with the goal to facilitate the provisioning of multipath connections over dynamically selected domain chains. In this method, we propose to virtualize the network into multiple slices and represent as multiple routing topologies by the PCEs. In this way, multiple routing planes are constructed to facilitate inter-domain routing. For both methods, we consider the buffer size as constraint for multipath routing due to the resulting differential delay. The results show that the proposed methods can significantly improve the blocking performance and are backward compatible, while only slightly impacting the intra-domain traffic.


Available from: Mohit Chamania, Sep 12, 2014
  • [Show abstract] [Hide abstract]
    ABSTRACT: Quality of Service-enabled applications and services rely on Traffic Engineering-based (TE) Label Switched Paths (LSP) established in core networks and controlled by the GMPLS control plane. Path computation process is crucial to achieve the desired TE objective. Its actual effectiveness depends on a number of factors. Mechanisms utilized to update topology and TE information, as well as the latency between path computation and resource reservation, which is typically distributed, may affect path computation efficiency. Moreover, TE visibility is limited in many network scenarios, such as multi-layer, multi-domain and multi-carrier networks, and it may negatively impact resource utilization. The Internet Engineering Task Force (IETF) has promoted the Path Computation Element (PCE) architecture, proposing a dedicated network entity devoted to path computation process. The PCE represents a flexible instrument to overcome visibility and distributed provisioning inefficiencies. Communications between path computation clients (PCC) and PCEs, realized through the PCE Protocol (PCEP), also enable inter-PCE communications offering an attractive way to perform TE-based path computation among cooperating PCEs in multi-layer/domain scenarios, while preserving scalability and confidentiality. This survey presents the state-of-the-art on the PCE architecture for GMPLS-controlled networks carried out by research and standardization community. In this work, packet (i.e., MPLS-TE and MPLS-TP) and wavelength/spectrum (i.e., WSON and SSON) switching capabilities are the considered technological platforms, in which the PCE is shown to achieve a number of evident benefits.
    IEEE Communications Surveys &amp Tutorials 01/2013; 15(4):1819-1841. DOI:10.1109/SURV.2013.011413.00087 · 6.49 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The Path Computation Element (PCE) and PCE communication Protocol (PCEP) have been standardized by the IETF. The scalability of PCE and its capability of constrained path computation make it especially attractive to carriers. Considering the latest extensions on related protocols, PCE has become the de-facto standard in connection service provisioning, especially in the carrier grade networks with strict QoS requirements. Whereas the PCE frameworks proposed to date have exclusively considered single path computation, the increasing demand for flexible services with high bandwidth requirement is driving PCE towards multipath techniques. This paper proposes the first implementation and protocol extensions of PCE with capability of multipath computation in line with PCE standards. We show that the overhead, i.e., signaling delay, caused by the multipath extensions in the PCEs is acceptable, while the blocking ratio of connection requests can be significantly reduced. We also suggest that the number of paths can be limited in the multipath routing algorithms to reduce the overhead caused by multipath extensions at a price of slight increase of blocking ratio. The results show that the presented multipath PCE system is scalable in large networks, with average signaling delay in the order of milliseconds. Therefore, it is feasible in practice and it can be beneficial for transport networks with emerging new applications, such as optical and Carrier Ethernet networks.
    Network and Optical Communications (NOC), 2013 18th European Conference on and Optical Cabling and Infrastructure (OC&i), 2013 8th Conference on; 01/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: Internet continues to suffer from limited bandwidth, highly varying delay, and losses. In literature, several studies suggested that an application can increase its aggregate throughput, and reduce its end-to-end delay and losses, by distributing data over multiple paths. In this work, a novel routing algorithm based on prediction method and optimization theory is proposed for improving the quality of service over the multipath network. The paper describes how the algorithm dynamically predicts loss levels over all available network paths and adjusts the traffic allocation proportion of packets for next prediction period by the PSOSA. A number of experiments are performed. The proposed algorithm makes the most promising data forwarding method with an effective network performance, even under unfavorable scenarios.
    2013 6th International Conference on Biomedical Engineering and Informatics (BMEI); 12/2013