With massive growth of data traffic, transport infrastructure needs to evolve to a packet-based transport paradigm in order to close the disparity between traffic and revenue growth. MPLS-TP, a profile of the widely-deployed IP/MPLS protocol suites, is an emergingdustry standard aimed at enabling connection-oriented packet transport.  MPLS-TP capitalizes on major IP/MPLS capabilities while introducing several additions that allow for operation and practices that are norm in the transport world.  This is crucial in order to enable a smooth transition from SONET/SDH-based transport to packet transport and preserve the operational benefits of efficient transport.  The existing IP/MPLS capabilities including  service and control plane scalability, resiliency, manageability, and traffic engineering with strong  QoS are  augmented in MPLS-TP in order to provide:

• enhanced OAM with additional capabilities for  performance management, multi-level operation, fast support for protection switching, various alarms etc.  The OAM is defined to be able to operate with or without the presence  of a control plane;
• separation of the data and control planes;
• static provisioning and operation through the network management plane  as well as dynamic establishment of connections through the control,  plane which is envisioned to be based on GMPLS.
• fast protection switching through the data plane, in addition to the control-plane-based fast re-route mechanism.

back to tutorials ^

MPLS/BGP Multicast Virtual Private Networks (MVPNs) have several architectural aspects and deployment characteristics. We examine three major independent components of MVPN technologies: discovery, multicast signaling, and core network transport. These components can be realized using different techniques and protocols. The possible MVPN deployment options, their benefits and tradeoffs are analyzed based on solution needs, scalability and convergence requirements, as well as MVPN characteristics in live networks. Experiences, trends, bottlenecks and scaling issues in existing large scale MVPN deployments will be shared to show that real scale issues exist elsewhere. Current knowledge of multicast traffic characteristics in VPNs will be shared and the importance of multicast traffic analysis will be discussed to help decide on the suited protocols

This tutorial will examine key aspects of MVPN technologies and compare different options for their realization to accommodate the needs of present and future networks. It will dispel the notion of “one-size-fits-all”, demonstrate why multiple solutions exist, and offer guidance on choosing the correct solution to address needs.

back to tutorials ^

MPLS has been used in Wide Area Networks, metro networks, Enterprise networks and Data Centers. We will see whether it is possible to extend MPLS to access networks, and the benefits and downsides of doing so. The current notions of MPLS networks -- the so-called "PE" and "P" routers -- will be challenged, and we will look at a more complete view of the network. We will then explore answers to the problems of service flexibility, service convergence and scaling. We will look at several approaches to bring MPLS LSPs to DSLAMs and other access equipment that may not have a full-blown MPLS control plane. We will tackle the overall problem of how to scale MPLS networks to 100,000 nodes, how to keep the control and data planes manageable, and how to achieve fast protection in an end-to-end fashion.

This tutorial is an updated version of last year's tutorial. The most important changes are: a discussion on the use of MPLS vs. MPLS-TP for in the access, and more technical details on scaling and protection.

back to tutorials ^

This will be available soon.

back to program ^

This will be available soon.

back to program ^

Use of LSP hierarchy provides a powerful mechanism for scaling MPLS infrastructure. However, none of the existing MPLS local protection mechanisms provide protection in the presence of LSP tail-end node failure - these mechanisms work only when failures occur at some intermediate nodes. In this talk we outlined a proposal for local protection of LSP tail-end node failure in the presence of LSP hierarchy.

back to program ^

This talk describes a means of scaling MPLS to very large networks, such as seen when migrating metro, access and mobile backhaul networks to MPLS. It first provides motivation for such a migration. Then, the talk describes some alternative mechanisms to achieve such scaling, followed by our proposal, and a discussion of the pros and cons of these mechanisms. This is followed by a deep dive into how the proposed mechanism works in detail. Finally, there is a description of how fast end-to-end restoration would work in such a scenario.

back to program ^

Using the "Seamless MPLS" approach Deutsche Telekom is currently building an MPLS network integrating 100.000 nodes spanning the Access, Aggregation and Core network. The goal is to provide End-to-End services for residential, whole-sales and business customers on a single MPLS platform. The presentation will give an insight of the Seamless MPLS architecture based on the existing requirements (scalability, redundancy, services, etc.) as well as and overview about the migration path from the existing network:

• Architecture Requirements
• Control and Forwarding Plane
• Scalability, Services, etc.
• Overall Network Architecture
• Access Network, Aggregation and Backbone
• Detailed description of MPLS features needed in the different areas of the network and how MPLS is used to fullfil the requirements
• Label on Demand for Label Assignment
• Fast Reroute in order to be below/around 50ms for Convergence
• Use of BGP in Seamless MPLS
• OAM ...
• Current state in standardization
• Roadmap and Migration from existing network towards Seamless MPLS
  where are we, what are the plans for the next years)
  
  

back to program ^

MPLS-TP rounds out the MPLS suite, enabling MPLS technology to provide end-to-end seamless networking. With MPLS as a unifying technology for the forwarding plane, the capabilities of any particular platform become a function of the control plane capabilities

This breaking down of technological barriers heralds a paradigm shift in how networks are designed, built and operated.

back to program ^

The IETF has standardized Ethernet pseudowires and VPLS, and these form a firm foundation for deploying a public Carrier Ethernet service. However, the IETF RFCs alone are insufficient to specify Verizon’s requirements for Carrier Ethernet over MPLS. In addition to the protocol mechanisms, Verizon has particular requirements related to conformance to MEF service definitions, reliability, control plane requirements such as convergence and interoperability, QoS, performance monitoring, OAM, manageability, testing, and certification. This talk will discuss the various aspects of these requirements in detail.

back to program ^

It was 10 years ago that Nasser El-Aawar, and I started the pseudowire design, in a coffee shop in Colorado. Today, Pseudowires based on the original draft-martini are one of the prevalent clients of MPLS networks to offer a wide variety of services. Recently there has been a new trend toward pseudowires without a control plane. Everything static! From pseudowire deployments in the Mobile RAN environment to SPs recently looking at offering a pseudowire interface to their customers, statically configured pseudowires are becoming very prominent. This trend is not particularly tied with the recent MPLS-TP technology evolution, in fact most of the applications use MPLS as the transport network. I will explore some use cases, applications of MS-PW to RAN, and recent enhancements to add pseudowire status support to static PWs and resolve some of the redundancy scaling problems.

back to program ^

Service Providers (SPs) often have strict SLAs (Service Level Agreements) for the services they provide to their customers. Often, an important requirement in a SLA is to provide an upper bound on the service unavailability time due to various failures within the service provider infrastructure (including the part of the infrastructure that connects customers' equipment to the service provider equipment). Moreover, this upper bound is required to be of the order of tens of milliseconds. That, in turn, requires fast service restoration with deterministic upper bound.

An important service offering by SPs today is VPLS. To provide service restoration with deterministic upper bound for VPLS service, SPs have to address three types of failures: (1) Access link failure, (2) PE node failure, and (3) P node or core link failure. There are no existing solutions today that provide fast service restoration with a deterministic upper bound for VPLS in case of type (1) and (2) failures.

For type (1) and (2) failures, existing solutions rely on control plane to converge before customer service is restored. Since control plane convergence is not deterministic, and thus, cannot be predicted beforehand, it is often not an acceptable solution for SPs who have to design their network to meet the SLAs for fast and deterministic service restoration with their customers.

The presentation describes a mechanism that provides fast and deterministic service restoration in case of type (1) and (2) failures.

back to program ^

This presentation provides a brief overview of Storage Area Network (SAN) services and how a MPLS or Carrier Ethernet network can be used to extend SAN services across a metropolitan area network (MEN) or wide area network (WAN). SAN over MPLS and Carrier Ethernet represents a new application to generate demand for MPLS/Carrier Ethernet services by business customers. This presentation will also discuss the service requirements of SAN services over a MPLS and Carrier Ethernet Network.

back to program ^

To manage and reduce the inherent complexity of large-scale networks, it is necessary to decompose the infrastructure into smaller building blocks or sub-networks and specify the interfaces and relationships between the sub-networks. Constituent sub-networks in turn can be recursively decomposed into smaller subsystems. Network architecture specifies the overall structure of the infrastructure in terms of sub-networks, subsystems, or building blocks and the interfaces and relationships between them. A model for describing the architecture of a network must offer multiple concurrent views of the infrastructure at different levels of abstraction to serve the needs of various stakeholders. Ultimately, network architecture emanates from a set of business and technical decisions and exerts influences on the network and the organization that spawned it long after the service life of individual network elements. This presentation highlights some architectural principles for the new public Internet.

back to program ^

The Path Computation Element (PCE) has become a familiar concept and multiple implementations exist to determine end-to-end paths in multi-domain MPLS-TP and MPLS-TE networks. PCEs can be used in the per-domain path computation technique to devolve to each domain entry point the responsibility to compute each path segment. Alternatively, the backward recursive path computation (BRPC) mechanism allows the PCEs to collaborate to select an optimal end-to-end path that crosses multiple domains.

These techniques, however, assume that the sequence of domains to be crossed from source to destination is well known. No explanation is given of how this sequence is generated or what criteria may be used for the selection between domains. In small clusters of domains, such as simple cooperation between adjacent ISPs, this selection process is not complex. In more advanced deployments (such as meshed-AS environments) the choice of domains in the end-to-end domain sequence can be critical to the determination of an optimum end-to-end path.

In this presentation we explain the latest mechanisms for the selection of domains that comprise the end-to-end domain sequence. We examine how BRPC could provide the necessary function, but show how it doesn’t scale to large networks. Instead, a new model of hierarchical PCEs is explained. This way of performing PCE cooperation enables rapid and scalable selection of optimal paths and can take advantage of both typical traffic engineering constraints (hop count, bandwidth, SRLG, path cost, etc.) and also more commercially relevant constraints such as policy, SLAs, security, peering preferences, and dollar costs.

back to program ^

This presentation addresses motivations for seamless interconnection of transport networks for end-to-end service. Furthermore, this presentation introduces “thin” layer concept as an enabler of seamless interconnection of transport networks, and discusses how recent technology development will fit within this vision.

Currently, various new transport networks are investigated, such as WSON (Wavelength Switched Optical Network), MPLS-TP, and PBB-TE. It is expected that these technologies will be deployed on a small scale at the start, and will co-exist. For example, MPLS-TP, PBB-TE may be deployed in the access/aggregation network, ROADM in the metro, and WDM in the core.

Here a challenge is how to interconnect these various transport networks for end-to-end services. Usually, this type of interconnection is not directly achieved at the transport layer. There are service networks on top of transport networks. For example, IP routers are placed at the boundary of aggregation/metro, and metro/core, and terminate transport paths for IP forwarding.

However, there are several benefits for seamless interconnection at the transport layer, such as:

• No need to place service specific nodes at every boundary of transport technologies. For example, early deployment of a service network may place service nodes at the border of metro/core, bypassing aggregation/metro boundary.
• Multiplexing among service networks is possible.

One potential solution is to deploy a single transport technology across all areas, eliminating the need for interconnecting different technologies. Although this may be ideal, there are several drawbacks, such as cost and migration barrier.

Therefore, it would be attractive to interconnect various transport networks, via a “thin” layer. The thin layer behaves as top of stack and interconnects various transport networks. Requirements from carrier perspective for such “thin” layer include support of wide range of underlying transport technologies, support of wide range of services on it, support of wide range of bandwidth and quality (e.g., delay, bandwidth) to match requirements of each service network, and manageability (e.g., provisioning speed and complexity, end-to-end monitoring).

Candidates for “thin” layer basis are Ethernet and MPLS/PW. On the Ethernet side, for example, MEF E-NNI is being specified as a technology to interconnect multiple domains at the Ethernet level. On the MPLS/PW side, for example, MS-PW is being specified as a technology to interconnect multiple domains at the PW level. This presentation analyzes technology development of relevant technologies, including control plane technologies such as GMPLS and PCE.

Finally, this presentation proposes future directions to address requirements from carrier perspective, including expectations for standardization.

back to program ^

back to program ^

MPLS-TP technology is gaining the service providers’ mind share. In this presentation, we discuss the reasons behind the major interests for MPLS-TP; transport transition happening; MPLS-TP characteristics; and the potential use case scenarios for MPLS-TP.

What problems are we trying to solve? With many new services being introduced by service providers, the demand for bandwidth has been increasing dramatically, and it is expected to accelerate for the years to come. Existing transport infrastructure is no longer meeting the needs to support the growing new services, including the mobility and triple play services. Service providers are looking for the next generation transport technologies which can provide higher bandwidth, efficient bandwidth utilization, and lower cost through statistical mixing rather than fixed bandwidth. The industry is moving away from SONET/SDH and TDM/ATM technologies to packet transport technologies, such as MPLS-TP, to provide efficient high bandwidth with low operational cost. – The transport world is at a turning point.

MPLS-TP is aimed to maintain transport characteristics through the technology transition, by providing pre-determined path, in-band OAM, fast detection and recover time, NMS provisioning or control plane options, tight SLA, bandwidth reservation, QoS, and high availability. The interoperability of MPLS-TP and MPLS makes the solution a good complimentary to the existing MPLS deployment. The transport flavored OAM and protection mechanism makes MPLS-TP a welcome choice by the transport world. Deployment scenarios for MPLS-TP include metro aggregation/access, Wireless backhaul, and multi-service Wireline transport. A new chapter is about to open.

back to program ^

This talk will describe a MPLS based architecture and solution for Service Provider (SP) metro and backbone transport networks. It will also provide a perspective on the applicability of the emerging MPLS-TP technology to transport networks.

It will start by describing the requirements as SPs transition to packet based transport networks, using ethernet, and seek greater integration between the optical and packet layers. It will then take a look at the capabilities available in the current MPLS/GMPLS technology to meet those requirements. It will also describe the perceived gaps in the current MPLS/GMPLS technologies with respect to certain requirements such as OAM. The relevance of such gaps in a packet based environment will be discussed along with solutions to address those gaps. The talk will describe the enhancements being made by MPLS-TP to MPLS technologies, such as MPLS OAM, and the implications of those enhancements.

back to program ^

Multicast services are increasingly in demand for high-capacity applications such as multicast VPNs, IPTV (on-demand or streaming), and content-rich media distribution (for example, software distribution, financial streaming, or data-sharing).

Operators of packet transport networks have been quick to recognise the benefits of multicast as a tool to optimise network resource usage, and to transmit only one copy of data on shared links towards multiple destinations.

Additionally, the use of MPLS-TP helps provide aggregation, Quality of Service (QoS) through Resource Reservation (RR), Traffic Engineering (TE), while providing Operation and Management (OAM) of the network. A Path Computation Element can provide the necessary path computation to optimise network resources in order to carry multicast services efficiently across the transport network, and resolve path requirements such as protection, load balancing and SLAs.

In a multi-layer network the benefits of IP multicast and P2MP MPLS cannot be fully realised without also considering the interactions between domains. For example, it is crucial to consider the domain sequences, domain policies and choice and number of domain interconnects.

This presentation will show how to select domain sequences for P2MP transport services. We will examine the trade-off between branching in an upstream or a downstream domain to achieve an optimal path versus administrative simplicity. The presentation will also show how the use of hierarchical P2MP tunnels to carry traffic across a domain is also a practical solution to an inter-layer network problem; in a packet-optical network it can be beneficial to utilise P2MP server layer trails to achieve the best use of resources across all network layers.

back to program ^

The use of RSVP P2MP LSP to forward multicast packets frees the core routers from complicated PIM configurations. It provides resilience using traditional FRR methods and allows the ability to steer packets over traffic engineered paths to manage bandwidth network wide.

There are a number of considerations associated with configuring and deploying such a solution. This co-presentation will discuss topics which include the following:

1. the mapping of multicast groups to a P2MP LSP at ingress and egress LER and BUD LSR.
2. dealing with S2L path re-merge in the network.
3. global P2MP LSP re-optimization versus individual S2L sub-LSP re-optimization.
4. Multicast source redundancy and fast failover of UMH (upstream Multicast Hop) in MVPN using RSVP P2MP LSP.

back to program ^

The ability to compute constrained Traffic Engineering Label Switched Paths (TE LSPs) for point-to-multipoint (P2MP) LSPs in Multiprotocol Label Switching (MPLS) and Generalized MPLS (GMPLS) networks across multiple domains (where a domain is a collection of network elements within a common sphere of address management or path computational responsibility such as an IGP area or an Autonomous Systems) has been identified as a key requirement. This presentation outlines a PCE based path computation method to find remerge free and better route for a P2MP LSP. Specifically, it addresses this requirement by extending backward recursive path computation (BRPC) technique proposed for Point-to-Point (P2P) LSPs in for P2MP LSP path computation in a multiple domains network. The presentation also discusses extensions to the Path Computation Element Communication Protocol (PCEP) for Point-to-Multipoint Traffic Engineering LSPs.

The presentation also outlines a framework and required protocol extensions needed for establishing and controlling P2MP MPLS and GMPLS TE LSPs in multi-domain networks. In this respect, the presentation outlines a signaling based method to find remerge free and better route for a P2MP LSP. The signaling based requires bear minimal signaling extension and yet perform a way to find remerge free and better route for a P2MP LSP in a multi-domain network.

back to program ^

The advent of consumer targeted smart phones and personal devices has accelerated the cell phone penetration rate. This is leading to a large increase in mobile broadband traffic.

Current industry trends indicate a serious evaluation of the convergence of fixed and mobile networks for the purposes of OPEX and CAPEX savings. The content of the World Wide Web which has been built over many years is based on the TCP/IP protocols. This naturally lends to the adoption of the IP technology in the mobile world. Mobile IP (v4 or v6) helps in leveraging the IP based network assets while addressing the mobility aspects. However it requires upgrade of networks and the mobile node software for the purposes of handling mobility of the mobile nodes.

With the convergence of fixed and mobile networks there is an opportunity to leverage some of the already existing technologies to handle the mobility in the network. With Ethernet being the Layer-2 technology of choice in the telecom industry, this presentation will evaluate the use of VPLS and VLL as a building block for mobility.

VPLS and VLL have matured as a technology over the years with the development of Pseudo Wire resiliency, redundancy, OA&M and the Auto-discovery features around the core functionality. Further the underlying MPLS technology offers many options for the purposes of transport resiliency like millisecond re-routing. Additionally VPLS and VLL based solution removes the burden of address management in the foreign agent network for the mobile node roaming case, thus turning the foreign agent configuration into a simple plug-n-play case. This presentation will focus on VPLS and VLL as a viable and necessary technology for mobile networks.

We would like to submit this summary towards “Category E” of the topics highlighted in the CFP webpage of MPLS 2009.

back to program ^

Mobile service providers are migrating from their current use of TDM or ATM circuits over SONET networks to IP/MPLS over Ethernet networks. One of the services that has been available from the SONET networks is the availability of highly accurate frequency for use by the mobile basestations. This service must now be met over the Ethernet network. There are two principle techniques available to deliver this service: layer 1 frequency distribution (i.e. Synchronous Ethernet) and packet based timestamp delivery (e.g. NTP or 1588). This talk will discuss both the requirements of the mobile basestations for accurate frequency and phase and discuss these new techniques that are evolving to meet these requirements.

back to program ^

The term “Mobile Backhaul” often means different things to different people. This presentation will analyze what these two words mean to the two key entities that rely upon each other to deliver wireless services to end users:

1. Wireless service providers
2. Wireline carriers providing transport services to the wireless service providers

Often times, these two entities are from either completely different companies or are independent groups within the same company, just managed and operated differently. These two entities fundamentally serve different purposes and it is important that a clear distinction be made with respect to what “Mobile Backhaul” means to each of them.

The presentation will identify the key transport/network requirements for these two entities and explain why these requirements are often times different. In this context, we will analyze the applicability of P2P-, P2MP- and MP2MP-based connectivity models to each of these two entities to be able to deliver 4G services based on the LTE 3GPP2 standards. We will examine some of the key Layer 2, Layer 2.5 and Layer 3 networking technologies that can deliver the connectivity model requirements for these two entities. The technologies to be considered in this presentation are MPLS (P2P and MP2MP/VPLS), native IP-based forwarding, native Ethernet bridging and Connection-oriented Ethernet (both native Ethernet-based approaches and MPLS-TP).

The presentation will explain why some of these technologies are better suited for mobile backhaul than others, within the scope of a particular entity. It will conclude by making a recommendation as to which of these technologies are best suited for each of these two entities based upon the previously discussed requirements. It will also explain how the recommended technologies within each of these two entities can work seamlessly, particularly from an OAM and protection perspective.

back to program ^

Wireless networks backhaul and core requirements are evolving rapidly, requiring more flexibility, higher reliability, support for advanced QoS, mobility and lower cost. Drops in revenue per bit in service providers’ business plan and scalability limits of current network architecture are challenges faced by wireless service providers. MPLS can provide an ultimate solution to improve overall service reliability, scalability, and QoS in 4G wireless networks such as WiMAX/LTE. In this presentation, we will provide the network architecture for 4G backhaul and core network, and address the challenges and solutions in designing such a network.

back to program ^

Deutsche Telekom has deployed MPLS in the backbone in 1999. With the installation of GbE/MPLS based aggregation network, Deutsche Telekom has set the cornerstone to extend the MPLS "islands" into one seamless MPLS platform. This Seamless MPLS approach should provide the required flexibility and operational opportunity given by MPLS protocols. The deployment of the Seamless MPLS approach is one major project within Deutsche Telekom, with main activities in harmonizing engineering, planing and operating between access, aggregation and backbone and between traditional transport and paket based services.

The Seamless MPLS approach has not only technology aspects. It has also a lot of impacts of the way how a traditional telco company with a complex organisation works. This has to be kept in mind when developing a Seamless MPLS approach. Thomas Beckhaus is the network architect of the Seamless MPLS approach. He is with Deutsche Telekom since many years and will give an internal view into the first experiences implementing the Seamless MPLS approach in a complex telco organisation.

back to program ^

As applications grow in complexity and telecommuting becomes more widespread, enterprises and government bodies must consider a variety of options when choosing WAN technology. MPLS and VPLS often emerge as the top choices, but neither technology is suitable for every implementation.

MPLS is a strong option for large, multi-national corporations that are managing hundreds, even thousands of remote sites across the globe. Operating at Layer 3, MPLS is a solid choice for those businesses that have migrated away from legacy services. Recently, VPLS has been gaining ground as an alternative to MPLS. Layer 2-based VPLS technology is better suited to businesses with fewer sites to manage, but richer applications traveling across their network. For bandwidth-intensive applications such as videoconferencing, storage, disaster recovery, and VoIP, VPLS delivers an ease of management that is crucial for IT managers. Additionally, a hybrid network design that deploys both MPLS and VPLS has been emerging as another viable option.

This session will provide the audience with a better understanding of when to implement either technology or when to use a hybrid approach. We are no longer speaking to the customer survey, hence I removed the reference to it.

back to program ^

Ethernet services have traditionally been provided over Fiber access, however, the usage of HFC access could become a cost-effective way to offer Business Ethernet services. Interesting enough, fiber penetration in many service providers markets is less than that of HFC penetration (some studies put the Fiber:HFC coverage ratio to 1:4) wrt commercial areas coverage. Many of those commercial businesses may already subscribe to Cable TV services and desire T1-like services.

As Cable SPs/MSOs look at extending their services into the Small & Medium Business (SMB) market and remote site of Enterprise networks, Ethernet L2VPN services over DOCSIS/HFC becomes one of the key components in their business services portfolio. The ability to gain a competitive advantage in the Ethernet market segment by leveraging their DOCSIS/HFC infrastructure while building out fiber to the premises (E-FTTP), is quite strategic for cable SPs/MSOs.

This presentation will describe a solution architecture and strategy for Business Services over DOCSIS/HFC and Fiber access in a converged IP/MPLS network environment. This presentation will primarily focus on designing & delivering MEF based L2VPN Ethernet services e.g. E-LINE and E-LAN and explain how MPLS capabilities are leveraged.

The presentation will detail the advantages & disadvantages of various design options, and also describe innovative solutions to address key operational issues in offering business services over DOCSIS. Lastly, it will discuss the interworking aspects for maintaining DOCSIS and Fiber access transparency (to ensure same experience to the VPN customer whether its sites are connected via DOCSIS or Fiber or both).

back to program ^

The control plane is separated from the data plane in GMPLS-based optical networks such as TDM, LSC, and FSC networks. Cross-connections in nodes on the label switched path (LSP) are made and/or released under control of the control plane. The key question in such networks is when it is safe to send the data if the LSP is established. Do we need additional mechanisms to verify the liveness of the LSP data-plane? Do we need to run a hello protocol in the client laye (for example, in the IP layer) to verify the liveness of the optical layer data-plane? Do we need to run OAM mechanisms to verify the optical layer data-plane liveness?

With the increasing deployment of GMPLS-based optical networks, LSP establishment performance is of fundamental concern. This issue is important when we interpret the performance of equipment being tested or considered for purchase and deployment. For example, the dynamic provisioning performance metrics set out in [2] need to be understood in this context ­ it is not sufficient simply to know how rapidly a node can transmit and process a control plane message; we must also know how quickly the data plane can be prepared for use.

In this talk we clarify the GMPLS signaling message exchanges in the light of data-plane liveness, and discuss whether we need additional mechanisms in addition to those already present in the GMPLS control plane to verify the data-plane liveness. We believe that this issue has to be solved for carrier-grade network operation, and note that it is now being discussed within the IETF [1].

[1] K. Shiomoto and A. Farrel, ³Advice on When It is Safe to Start Sending Data on Label Switched Paths Established Using RSVP-TE,² draft-shiomoto-ccamp-switch-programming-00.txt, 2009 March.

[2] Sun, W., and Zhang, G., "Label Switched Path (LSP) Dynamic Provisioning Performance Metrics in Generalized MPLS Networks", draft-ietf-ccamp-lsp-dppm, work in progress.

back to program ^

We implemented Point-to-Multi-Point (P2MP) path provision supported by GMPLS control on the experimental Ethernet network, assumed by Wide Area Ethernet. In this implementation, GMPLS RSVP-TE is extended for establishing P2MP Label Switched Paths (LSPs), based on the standard of RFC 4875. The signaling scheme to establish P2MP paths is reported together with the experimental results.

This presentation contributes directly to the development of the future of Wide Area Ethernet. Ethernet has greatly succeeded in Local Area Network (LAN). Its simple transmission scheme facilitated the development of high-speed and cost-effective equipments. By the achievement of Gigabit-class interface and stacked Virtual LAN technology, Ethernet is now expanding to carrier class transport network as Wide Area Network (WAN), which is called Wide Area Ethernet.

Its transmission speed has reached to 10Gbps with IEEE 802.3ae in 2002, and 100Gbps Ethernet will be standardized with IEEE 802.3ba by 2010. Ethernet which obtains these high-speed technologies, it is replacing legacy transport network technologies such as SONET/SDH. However, current Ethernet technology has critical problems to become Wide Area Ethernet on its control plane. The network scalability is restricted due to characteristic MAC learning and forwarding system, and spanning-tree protocol (STP) provides unreliable restoration scheme that takes long convergence time in a larger size network. In addition, current Ethernet does not support advanced Traffic Engineering (TE) or the deliver of Quality of Service (QoS), which both are required for Wide Area Ethernet.

GMPLS is a strong candidate to solve these problems. It provides an advanced control plane on any network layers. In GMPLS framework, the control plane is divided from the data plane and establishes connection-oriented paths in the data plane. This control plane will provide reliable restoration or protection, advanced TE and the deliver of QoS, which will satisfy the requirement for future transport networks. To achieve Wide Area Ethernet as a transport network, it is required to extend GMPLS for supporting Ethernet. IETF CCAMP and GMPLS controlled Ethernet Label Switching (GELS) are standardizing these extensions.

Wide Area Ethernet, which gains GMPLS control plane, will next require definitely the P2MP path provision. Its Gigabit-class high-speed interface makes the origin from electrical switching system, which can easily duplicate incoming data traffics to all switch interfaces adapted for providing P2MP paths. Especially for Wide Area Ethernet, the demand for P2MP Layer-2 VPN service is quite large. Therefore, it is a very important challenge to establish a P2MP path in Wide Area Ethernet. In this implementation, we achieved P2MP path provision by connection-oriented GMPLS controlling.

GMPLS achieved a standard for extensions to RSVP-TE for establishing P2MP TE Label Switched Paths (LSPs), RFC 4875. We extended RSVP-TE based on this RFC 4875, and achieved P2MP path provision on experimental network assumed by Wide Area Ethernet. The experimental network is including the data plane, and the control scheme for the data plane is also reported.

back to program ^

For the next generation wide area networking, we proposed the virtualized network using the MPLS-TP (Multi-Protocol Label Switching-Transport Profile) technologies and developed a packet transport system. With this packet-base transport technology, many kinds of networks, for example, leased line and carrier Ethernet can be converged. In addition to it, this network can be used for mobile backhaul and Optical access network.

1. Backgrounds
   
   As network application used in business scene, importance of networking for business is expanding. The customers choose the network which fits their needs. The biggest two requirements are “wide band width” and “reliability”. For these two requirements, two kinds of network are used. One is leased line or telephone network based on transport technologies like SDH (Synchronous Digital Hierarchy), which is very reliable but relatively slow. The other is packet-based VPN (Virtual Private Network: IP VPN and Ethernet VPN) services based on best effort IP and Ethernet technologies, which is very fast and best-effort basis. Thus, network operators have many kinds of networks to meet these requirements.
   
   Our challenge is to meet these two requirements on one network. For that purpose, we developed the packet-based reliable network with MPLS-TP technologies.
   
   
2. Concept of the virtualized network
   
   We propose the next generation virtualized network, with MPLS-TP (Multi-Protocol Label Switching-Transport Profile) technologies, which are standardized at IETF (Internet Engineering Task Force). The valuable length packet technologies are advantageous to make high speed network. Then we introduces static MPLS-TP tunnel virtual path which is managed from NMS (Network Management System) and each virtualized network accomplishes both wide bandwidth and high reliability. To make the network reliable, OAM (Operation Administration and Maintenance) and protection tools are implemented on each virtualized network. For assured transfer, all the paths are static and controlled with NMS.
   
   
3. Implementation of reliable transfer mechanisms on the system
   
   It supports the OAM functionality for each LSPs (users), continuity verification, AIS/RDI, and Loopback. As all the technologies are implemented as hardware, these functions can be performed respectively. For all the LSPs, 1:1 protection with APS (Automatic Protection Switching) mechanism is implemented.
   
   For assured transfer, all the paths are static and controlled with NMS and when a path is set up, each node on the path calculates not to exceed the packet handling capacity at the node. In addition to it, at the entrance there is packet shaper for each path to control the traffic.
   
   
4. Adaptation with leased-line network
   
   Not only packet based network, but traditional leased line or other networks are connected to this virtualized network. For that purpose, we developed the circuit emulation function for ATM interface. Each ATM virtual path is connected to MPLS-TP virtual path and functions are interconnected each other.
   
   
5. Conclusion
   
   We developed the next generation packet transport system, which is very reliable and wide bandwidth network. In the system, we used static path with MPLS label set by NMS and supervise these paths with embedded OAM tools. This network can adopt both leased-line and packet-base VPN.

back to program ^

Ethernet services have traditionally been provided over Fiber access, however, the usage of HFC access could become a cost-effective way to offer Business Ethernet services. Interesting enough, fiber penetration in many service providers markets is less than that of HFC penetration (some studies put the Fiber:HFC coverage ratio to 1:4) wrt commercial areas coverage. Many of those commercial businesses may already subscribe to Cable TV services and desire T1-like services.

As Cable SPs/MSOs look at extending their services into the Small & Medium Business (SMB) market and remote site of Enterprise networks, Ethernet L2VPN services over DOCSIS/HFC becomes one of the key components in their business services portfolio. The ability to gain a competitive advantage in the Ethernet market segment by leveraging their DOCSIS/HFC infrastructure while building out fiber to the premises (E-FTTP), is quite strategic for cable SPs/MSOs.

This presentation will describe a solution architecture and strategy for Business Services over DOCSIS/HFC and Fiber access in a converged IP/MPLS network environment. This presentation will primarily focus on designing & delivering MEF based L2VPN Ethernet services e.g. E-LINE and E-LAN and explain how MPLS capabilities are leveraged.

The presentation will detail the advantages & disadvantages of various design options, and also describe innovative solutions to address key operational issues in offering business services over DOCSIS. Lastly, it will discuss the interworking aspects for maintaining DOCSIS and Fiber access transparency (to ensure same experience to the VPN customer whether its sites are connected via DOCSIS or Fiber or both).

back to program ^

In the past there has been much controversy between IPFRR and MPLS. Too often debaters have been assuming that IPFRR and MPLS offer disjoint solution sets for service restoration.

This talk is about explaining a protocol architecture which combines the best of two worlds. The necessary interfaces between the OSPF and IS-IS link-state protocols and the MPLS signaling protocols (most notably LDP and RSVP)will be highlighted by example of a real world implementation.

Issues like CPU computation scalability for complex (node) protection scenarios, as well as how to achieve 100% backup coverage will be discussed - finally a novel solution, which fits nicely into the MPLS framework will be presented.

The talk will be complemented by an assessement of how to deploy IP/LDP FRR to make it scale into the largest networks in the world, including some testing results demonstrating that 50ms service restoration is feasible today.

back to program ^

Next-gen L3VPN deployments require fast convergent, scalable and stable design as end customers are moving their time critical data traffic to shared L3VPN fabric of service providers. These 3 goals sometimes get into conflict. This paper talks about evolution from scratch of Alcatel 7750 L3VPN software architecture blocks that blend these 3 design goals. Fast convergence is achieved through loose binding of different control plane blocks (mpls, routing, forwarding) in data plane and fast next-hop tracking (prefix to nexthop indirection which Alcatel 7750 was first to introduce in 2004). Scalability is achieved through various interesting techniques like reusable elegant design blocks that are optimized through multiple indices to the same database (e.g. RIB), data structures optimized for faster access etc. Alcatel 7750 router was the first in the industry to offer High Availability (HA) L3VPNs through stateful syncing of information to standby control card, which is one step beyond stateful failover of routing protocols so other nodes in the network are agnostic of nodal failure. This talk also talks of how these 3 different goals are meshed together and concludes with recommendations for deploying L3VPNs for highly scalable, convergent and stable network.

back to program ^

Enterprise networking solutions are composed of complex service offering that change dynamically and are configured uniquely for each customer. Consequently, the sales and implementation processes are lengthy and inefficient and customer proposals are inaccurate, often resulting in high fulfillment rejection rates.

An integrated collaborative solution design platform that can handle the complex enterprise network model, complete service and product offering rules, as well as a process workflow engine, can significantly improve sales productivity, increase customer satisfaction and reduce rejection rates.

A case study will illustrate how one global communications service provider overcame these challenges successfully.

back to program ^

As networks become increasingly faster and most complex, service providers are finding themselves continuously searching for more efficient OAM tools in order to maintain customer satisfaction. In fact, many of these customers are not just hoping for but demanding that providers include stronger service level agreements for Ethernet and IP/VPN services.

Being able to deliver strong SLAs is harder than it initially seems – complexities throughout the network can inhibit successful end-to-end delivery. Without thorough, active, and measureable network testing equipment, service providers would be signing their customers up for ineffective SLAs, resulting in unreliable service and customer churn.

Atul Bhatnagar, president and CEO of Ixia, will discuss what the primary characteristics of a disrupted network system are and how service providers can ensure service providers they are promoting strong SLAs for their customers. Mr. Bhatnagar will explain the current state of IP/MPLS network testing and what components are required for a truly comprehensive network test, such as being able to deliver end-to-end active monitoring and analysis of multiplay networks. He will also discuss the pertinence of testing equipment placement – being able to isolate and test problems at different points throughout the network segment – a commonly overlooked testing feature.

back to program ^

The introduction of Triple Play services such as IPTV saw a new and exciting time for consumers. As MPLS expands out of the Core and into the Access, network equipment manufacturers and service providers must ensure that new MPLS devices and new MPLS-based networks continue to meet scalability, performance and reliability expectations for Triple Play services, including IPTV.

In this presentation we will discuss mechanisms used to support Triple Play services over MPLS in the Access, including mechanisms for efficient multicast service delivery, such as Point-to-Multipoint (P2MP) and Next Generation Multicast VPNs (NG-mVPNs) and Multicast VPLS. We will describe the test methodologies involved in validating the interoperability, scalability and performance of devices and networks implementing these approaches. You will also be presented with key test scenarios to evaluate and measure the impact they have on the end user experience.

back to program ^

With the rapid growth of internet traffic and the squeeze on financial budgets, many carriers have moved from a position of excess capacity to one where careful planning decisions have to be made. Existing capacity needs to be used effectively; new capacity needs to be added when and where it is most needed. At the same time, networks still need to be resilient enough to survive failures with reasonable capacity.

Direct measurements of end-to-end traffic in IP/MPLS networks are difficult and at best only cover part of the network, such as the MPLS core of a network with a full mesh of traffic engineering tunnels, for which traffic can be measured. In most circumstances, in order to build a more realistic traffic model, we need to make use of detailed customer knowledge of the traffic distributions in addition to limited measurements of interface and/or end-to-end traffic.

In this talk, we present a number of customer case studies with mobile carriers, internet service providers, and global network operators. Here we work with the available network measurement information combined with user-intelligence to produce good quality models of the traffic. Crucial to this process is the ability of the software to model the traffic routing through the network accurately, whether based on ISIS or OSPF protocols and areas. MPLS-TE tunnel paths add further complexity to the routing.

As long as this routing is well understood, the traffic model we generate is ideal for further planning tasks for the customer. From this point, the customer can assess the impact of failures on this traffic, determine the capacity bottlenecks that may result from OSPF/ISIS metric changes, and make well-founded decisions on where to deploy or remove capacity.

back to program ^

GMPLS provides powerful tools for the control of optical networks. However, it does not provide ways of adapting local configurations in regards to global business-level goals. Label Switched Paths can be setup all at once, thanks to efficient signaling protocols, but the network configuration still has to be constructed manually and off-line. Although several propositions have emerged to enable dynamic reconfiguration (Path Computing Element architecture) and automate some parts of the decision process (Policy Based Management), none of them was able to adapt the network configuration properly to changes in the environment (traffic matrix, topology changes). During the last few years, a new paradigm has been brought into the focus of many research labs, under the topic “autonomic networking” also known as “self-management”. The goal is to let systems and networks carry some activities for their own management, thus enabling more on-line adaptive re-configuration and decreasing the overall management cost. In this paper, we demonstrate how a distributed platform for network self-management can be leveraged to automate the network configuration setup and thus can decrease the operation cost of GMPLS networks. For this purpose, we rely on one self-management platform, developed by Ginkgo-Networks, which we configure with proper behaviors. Our experiments show that in many cases, a GMPLS network properly managed by a distributed selfmanagement platform could do a better job (handle more traffic with less resource and reject less connections) than a statically pre-setup GMPLS network.

back to program ^

This abstract will be available soon.

back to program ^