Abstract:We analyze the objective cost of SBVTs as a function of the traffic conveyed by IP/MPLS-over-flexgrid networks. Optimal results show that cost increment with respect to BVTs is related to the aggregation level.
IntroductionNational IP/MPLS networks receive IP traffic from access networks and perform aggregation and routing. To minimize the number of ports a router hierarchy consisting of edge routers performing aggregation and transit routers providing routing flexibility is typically created. In addition, a reduced number of interconnection routers support inter-operator connections. For efficiency and reach reasons, IP/MPLS networks are usually designed on top of optical networks [1]. The maturity of the flexgrid technology [2] providing fine spectral granularity makes bandwidth-variable optical cross-connects (BV-OXCs) based on that technology the candidates to be deployed in operators networks in the near future ( Fig. 1).To take advantage from the mix of optical connection (lightpath) bitrates that can be created on flexgrid networks, bandwidth-variable optical transponders (BVT) can be used. The main advantages of BVTs with respect fixed transponders (FTP) are: i) the used spectrum width is a function of the actual lightpath's bitrate and ii) procurement and inventory is simplified, thus reducing operational expenditures (OPEX). However, when the associated lightpath's bitrate is low, BVT's capacity is wasted, thus increasing capital expenditures (CAPEX). For instance, imagine a 400Gb/s BVT currently used for a 100Gb/s lightpath. Notwithstanding, to ensure that the network can support the traffic forecast for the years to come, deploying transponders with lower capacity might also involve increasing CAPEX to replace those transponders in the near future. The answer to that dilemma is sharing each BVT among several lightpaths, as proposed in [3]. A sliceable bandwidth-variable optical transponder (SBVT) is thus a BVT that can be shared by multiple (f) lightpaths.We assume that IP/MPLS routers are connected to wavelength selective switches (WSS) in the local BV-OXC, similarly to the architecture in [4]. Virtual topologies are created by connecting IP/MPLS routers among them through lightpaths established in the flexgrid optical layer. IP/MPLS routers are dimensioned considering two main parameters: forwarding capacity and number of slots available in the chassis, which limits the number of cards that can be installed. Since the bitrate of the MPLS flows resulting from aggregating IP client flows at edge routers is usually low, those routers are connected to few transit routers and BVTs or FTPs are usually preferred. However, the connectivity of transit routers among other transit and interconnection routers is usually high, creating an almost full-mesh virtual topology among them (Fig. 2). Therefore, SBVTs could provide advantages by reducing the number of cards to be installed on transit and interconnection routers, thus saving CAPEX since cheaper routers could need to be installed.In this paper, ...