Carrier Ethernet is increasingly deployed in Metro and Access Networks. Vendors and Carriers alike have figured out how to scale Ethernet and how to deploy native Ethernet in the Metro region. Carrying Ethernet services across wide area requires integration of Carrier Ethernet technologies with prevalent and most dominant technology – MPLS. While it would seem trivial to integrate VLANs with MPLS using PWs, thanks to L2VPN work in the IETF, integrating a whole host of Carrier Ethernet technologies with MPLS PWs is not so trivial. This tutorial explores in depth the integration of 802.1ad, 802.1ah and Ethernet OAM with VPLS, MPLS VPNs and PW services. This tutorial also discusses topics such as redundancy and resilience, Ethernet as a Transport and provides a brief overview of latest developments wrt Trill, MPLS TP and 802.1aq.

back to tutorials ^

Widely deployed broadband is coming to mobile networks in the form of High Speed Packet Access (HSPA), Evolution Data-Optimized (EV-DO) and WiMAX networks to enable new mobile applications and services. Some of the biggest challenges that will result from this data and bandwidth explosion will be in the backhaul and aggregation functions that take place in the RAN. Historically RAN backhaul networks have been voice centric, low capacity and a source of major operational expense for mobile operators. New infrastructure technologies need to be able to support legacy access types, leverage cost points of newer transport types such as Carrier (Metro) Ethernet and leverage the high availability, resiliency, and Quality of Service capabilities of IP/MPLS technology.

This tutorial will examine the principal drivers for a new IP/MPLS backhaul transport infrastructure that accommodates the scaling needs of the evolving mobile networks. Key challenges, options, benefits and tradeoffs are explored for solutions supporting several prevalent applications. Key existing and emerging industry standards/agreements are referenced.

back to tutorials ^

MPLS has proved a very flexible and scalable infrastructure for service delivery and convergence, and is now the protocol of choice in Wide-Area Networks, and increasingly so in metropolitan networks as well. The next challenge is access networks and technologies: does MPLS have a role to play here? If so, what, why and how?

This tutorial addresses these questions. It begins with a discussion of the various types of access technologies, what functions access nodes serve, and how they fit in with the rest of the network. Then it discusses the benefits that MPLS can bring to access networks, and the requirements that "MPLS to the access node" must satisfy. The tutorial then describes an architecture for the control and data planes of access nodes of different flavors as well as services, and how the overall MPLS network can scale to 10s of thousands of nodes. Finally, it closes with a perspective on network and service convergence.

back to tutorials ^

This tutorial gives an all-round view of PCE from the architectural building blocks to the messages and object formats of PCEP. PCE existence and capability discovery will be explained. It describes the different PCE models that can be applied for computing individual or multiple paths, shows how stateless or stateful PCEs may be used, and explains how PCEs can cooperate to compute paths.

The key mechanisms of use and applicability of PCE will be summarised. The use of PCE to compute inter-domain paths through the per-domain and backward recursive path computation techniques will be explained, and the application of PCE to multi-layer networking for Virtual Network Topologies and the ASON architecture will be described.

Equally important, the tutorial will discuss the limitations of PCE, when it is not appropriate, and what work is still required to make PCE a fully flexible tool within MPLS-enabled networks.

back to tutorials ^

This will be available soon.

back to program ^

This will be available soon.

back to program ^

This presentation examines the key ideas behind BGP/MPLS VPN technology, the foundation for delivering scalable IP-based services profitably and shows how new ways of considering problems can result in innovative solutions. This session provides unique insights into the design and development of one of today's most innovative and successful technologies.

back to program ^

Organizations that have developed and operated "transport networks" a few years ago started to look for packet based alternatives to the circuit switched technologies. The efforts converged on MPLS and Transport MPLS (T-MPLS) were started to be designed.

The T-MPLS definition was less than optimal and turned out to be incompatible with MPLS as defined by the IETF. This was viewed as a very serious issue since MPLS is very widely deployed.

The presentation will give an overview of requirements and motivations for an MPLS Transport Profile, why the early efforts failed and high light the technical and organizational agreements that were reached in the discussions between ITU-T Study Groups and IETF Working Groups and Areas, as well as plans for how the issue will be managed in the standard organizations.

back to program ^

This presentation will discuss the current status and future trajectory of the introduction of a transport profile for MPLS technology. In addition, what will be explained will be the generic nature of the technology so that it will be useful for MPLS, Pseudowire and IP/MPLS networks. New features and functionality being introduced will be explained as well both with regards to transport networks as well as IP/MPLS services.

back to program ^

With the explosive growth of the Internet and, particularly in Japan, the expansion of low-priced broadband lines, the industry is moving ahead with enterprise service configurations, such as SaaS, that effectively utilize inexpensive broadband lines. At the same time, the enterprise service market is shifting from ATM/STM to Ethernet modalities. Given this environment, we have been studying; particularly in the area of next-generation enterprise services; next-generation service aggregation methods that replace ATM and that permit smooth migration with current service equipment. In this presentation, we will propose an architecture for traffic aggregation networks based on T-MPLS to provide next-generation enterprise network services.

NTT Communications Japan networks are formed from access networks, aggregation networks, core networks, and service nodes. NTT Communications is now constructing a nationwide-scale network. Aggregation networks connect core networks and access networks and, as such, exist to ensure highly efficient transport by bundling together multiple service access lines. Aggregation networks require resiliency on par with core networks since they accommodate large numbers of users and services.

We hope to employ T-MPLS to replace ATM in our future aggregation networks. Our reasons are as follows:

1. With T-MPLS, reliable path management is possible without control plane stipulations, although explicit paths must be set at other systems. And because the control and data planes are separated, simpler implementation of in-service software upgrades is believed possible, which will make it easier to realize high reliability standards.
2. OAM functions (MPLS-OAM/EtherOAM) equivalent to ATM OAM functions are being standardized, so that the ATM-level maintenance standards we require can be attained (LB/CC/PM/DM, etc.).
3. T-MPLS can efficiently accommodate existing services and provide resilient enterprise network services over broadband.

We will also discuss issues incident to migrating from fixed to variable packet lengths; bandwidth assurance methods, jitter, and regulating packet loss rates and other QoS parameters; when providing existing services that were originally built with ATM technology with Ether/MPLS-based technology.

back to program ^

With the emerging Triple Play Services highly scalable IP-Multicast solutions are needed especially in the aggregation networks in order to support aggregation domains with several 10.000 customers. In addition the use of IP-Multicast results in new security features which are necessary to protect the aggregation domains.

The presentation will give a detailed description of the Ethernet/MPLS aggregation network of Deutsche Telekom focussing on the multicast implementation for the Triple Play services. It describes the features and scalability parameters of the different network components in order to provide a very large number of customers with IPTV services based on IP-Multicast.

The following topics and questions will be covered in detail:

• Scalability parameters: Description of the parameters which are relevant for the scalability of an aggregation domain. This includes the number of concurrent users, number of IPTV channels (total, per user), bandwidth calculation for aggregation domains and the sizes of Multicast related tables (e.g. snooping database, forwarding tables, etc.) as well as QoS parameters and performance aspects. This takes also the evolution of the existing services into account to provide a view on future scalability parameters.
• IGMP and multicast features: Derives from the scalability parameters the features which are needed on the control and data plane for IP-Multicast. Based on the Ethernet and MPLS options for implementing aggregation domains an in depth description of the different solutions is given and the solutions are compared. The architecture for highly scalable multicast support in aggregation networks is depicted. Relevant security features to protect the platform and the implementation are described.
• Evolution and Multicast features for Ethernet/MPLS: Description of the current limitation of Multicast solutions for MPLS and Ethernet based networks which need to be addressed in the future.

back to program ^

MPLS/BGP VPN customers, i.e., enterprises and firms typically want to segment their VPN's by data center or hub location in order to meet specific performance, security, or application access requirements. Such partitioning of VPN's has been possible for unicast transmission. Multicast traffic in a VPN should be subject to similar partitioning by multicast source or Rendezvous Point location. Lack of such solution can deteriorate application's performance, introduce latencies and variable delays that impact users and applications. Typically, large enterprises have multiple data centers and would like multicast applications to be simultaneously sourced from each data center to serve different sets of branch locations. Locating sources close to receivers and partitioning of receivers by the source location is often necessary in business critical real-time data feeds.

We present Multicast MPLS/BGP VPN solution that allow the same multicast stream to flow simultaneously on multiple trees across provider's network without duplicate packets sent to customer receivers. The solution allows for different downstream VRF’s to choose different upstream PE's to customer RP or customer source. As a result, customer multicast stream can flow along multiple inter-PE trees and simultaneously utilize multiple paths in redundant topology. When preventing duplicates to customer receivers, the proposed solution does not waste provider's network resources by discarding, at network egress, already transmitted duplicate traffic. Only a single copy of any customer multicast stream will reach any egress VRF in a converged network. This includes PIM-SM streams that are either flowing on shared tree or on the shortest-path tree. An egress PE receives a PIM-SM stream either from the customer RP or directly from the customer source but never from both. The support of "anycast" sourcing includes Anycast RP operation by allowing multiple RP's to send traffic in parallel to their closest receivers.

In PIM-Bidir the partitioning of Multicast VPN by customer RPA location avoids multicast packet loops during routing convergence.

back to program ^

The dramatic growth of the internet routing table has an adverse effect on the dynamicity and churn of BGP routing, which in turn threatens control plane stability. This is making service providers rethink their strategy and consider removing BGP from the core and limiting its use to the provider edge. At the same time, service providers are rolling out internet video and IPTV services, requiring multicast distribution for efficient use of bandwidth. The use of MPLS P2MP does not require BGP and provides additional advantages over IP multicast with PIM in terms of resiliency and traffic engineering in the core.

This presentation gives an overview of engineering requirements of deploying a BGP-free core with MPLS P2MP for delivering Triple-Play services. Simulation results with OPNET SP Guru Network Planner provide visibility in the implications on the network performance, as well as the effects of optimizing traffic engineering for MPLS P2MP.

back to program ^

The first part will address Traffic Engineering as an instrument for optimizing the operational performance of networks and related core problems, discuss solutions in both connectionless and signaled approach, and point to topics for future research and development. Techniques such as multi-path routing, traffic splitting, constraint-based routing (CBR), path-protection etc. that are used for traffic engineering in contemporary Internet Service Provider (ISP) networks are discussed for both connection oriented ( e.g. MPLS ) and connectionless classes ( e.g., distance-vector and link-state algorithms, routing metrics etc.).

The second part addresses traffic engineering aspects of MPLS for Satellite Networks. The large-area coverage at efficient infrastructure costs makes satellite networks attractive from a provider’s point of view. However, communication suffers in case of geostationary (GEO) satellites from high free space attenuation, latency and link budgeting criteria. In contrast, Low Earth Orbit (LEO) satellites with their lower latency and link budgets and less free space loss offers better promises for a high efficiency solution. Advantages of MPLS such as forwarding based on exact match of fixed length labels, integration of ATM and IP technologies, traffic engineering and QoS routing, conveniently separated routing and forwarding networks etc. suggests that it can implemented over satellite links to guarantee required QoS service levels. In the past Asynchronous Transfer Mode (ATM) have been considered for dynamic or fixed satellite network topologies for reliable and robust traffic engineering methods. Connectionless IP protocol demands implementation of IP Quality of Service (QoS) architecture [Integrated Service (IntServ) or Differentiated Service (DiffServ)] to provide the required QoS, but does not satisfy mechanisms for traffic engineering as compared ATM. Multiprotocol Label Switching (MPLS), on the other hand, allows adoption of new parameters for conventional IP traffic by decoupling packet forwarding from the information carried in the IP header. This section discusses proposals for IntServ, DiffServ, and MPLS based new QoS architectures for satellite based IP networks.

back to program ^

We have discussed the basic concept of combing MPLS and IEEE 802.1ah to provide the Carrier Ethernet services last year. In this talk, we will present the practical deployment considerations and technical solutions for MPLS + 802.1ah. The discussion will focus on the following:

• Quick overview of MPLS + 802.1ah technologies and use cases
• H-VPLS extension with 802.1ah at uPE/nPE
• 802.1ah aggregation over MPLS core
• Design to support the scalable Carrier Ethernet services
• Migration technologies for partial 802.1ah network deployment
• Redundancy strategies
• OAM solutions

back to program ^

Interconnection of Internet Service Provider (ISP) networks, commonly referred to as “peering”, is a business arrangement between two companies whereby they agree to exchange Internet traffic between their respective customers. In the simplest case, connectivity is provided only between those sets of customers, with no transit service to third party networks, and no exchange of cash settlements; there are variants on this, such as “paid peering” and “partial transit”, which modify those parameters.

Peering is an important business function that can improve operational efficiency.

With peering agreements in place and well-engineered connections, an ISP can gain two advantages: reduced need to purchase transit and reduced latency/delay to destinations reached via peering connections.

To gain as much efficiency as possible from peering relationships, ISPs need to become effective at doing three things, which are the key success factors for peering.

1. Identifying the right peering partner(s);
2. Negotiating the best peering arrangements and terms for both parties; and 3. Establishing the most effective peering methodology and technical design configurations between the two providers.

Peering to date has been rather static between two parties.

In the context of MPLS, specifically RFC 4364, section 10 describes various mechanisms used to connect two VPNS sites that may span different multi-autonomous system domains, often referred to as to options a, b and c.

Service providers are implementing some of these mechanisms today.

However, service providers are asking the question as to scalability of this type of arrangement, and exploring the possibility of having 2 or more providers in a path e.g. multi-provider QoS constructs or service contractually signaling, all coming under so called Service Oriented Architecture <SOA> for interconnection.

What are these SOA-based attributes that comprise interconnection per se?

Additionally, large enterprise organizations also have their own sets of interconnect requirements that may use MPLS-based mechanisms.

The presenter will discuss SOA-based attributes that are emerging in the industry and additionally explore the following topics as part of the discussion:

• Scaling; security; topology hiding; OAM;
• Regulatory impact;
• Emerging mechanisms around service oriented architecture constructs like those being advocated by forums like IPsphere;TM Forum, OASIS,...
• Emerging mechanisms in SDOs like the IETF; ITU-T...
• Overall business impact

back to program ^

Next-Generation (NGEN) MVPN technology based on the aggregated routing architecture, BGP control plane, and MPLS data plane has matured in the last couple of years. This presentation describes the key benefits of NGEN MVPN, its scaling characteristics, and addresses concerns that have been expressed regarding this technology. This presentation provides an update on NGEN MVPN related standards activity in the IETF. It also provides an update on NGEN MVPN implementation and deployment, as well as its use as an infrastructure for other applications.

back to program ^

MPLS Traffic Engineering uses the RSVP-TE signaling protocol specified in RFC 3209 to install and manage traffic engineered LSPs in an MPLS network. RSVP-TE for GMPLS is specified in RFC 3473 and a series of subsequent RFCs and can be used to control a variety of networks such as WDM and TDM as well as MPLS packet switching networks.

The GMPLS control plane includes a set of useful features that are not available in the MPLS-TE specifications. These include the ability to set up bidirectional LSPs, the availability of prioritized and protected resources, and procedures for graceful restart. Perhaps equally important, the use of GMPLS protocols offers the chance to provide closer integration between different technology layers in the network with only one protocol implementation to diagnose and fix.

However, RFC 3209 and RFC 3473 implementations are not interoperable. Some of the fundamental objects that they use are incompatible, and the presence of the GMPLS objects is indicative of the mandatory support of a set of GMPLS functions while the use of the MPLS-TE options presumes that these advanced functions will not be available.

This presentation will examine the increasing demands to utilize GMPLS signaling functionality in MPLS networks from the requirement for bidirectional tunnels in Transport MPLS networking, to the closer integration of packet networks with the underlying transport networks. The speaker will show how these operational requirements are putting pressure on vendors to "leak" function from GMPLS into their MPLS-TE implementations, and how this threatens the interoperability of MPLS equipment.

The IETF has recently looked at this problem and proposed several ways to achieve the full GMPLS function set in MPLS networks. The presenter will express his opinion that a rapid migration to GMPLS signaling will be beneficial for vendors and carriers, and will make suggestions for how this can be achieved. The rapid developments of Transport MPLS and PBB-TE may used to leverage changes in the market and make the move to GMPLS less daunting.

back to program ^

MPLS is being extended deeper into the access and terminated at lower functioning devices. Many technologies exist and/or are being developed in this space. These include Pseudowires, Multi-segment Pseudowires and MPLS-TP. Further various means of signaling and configuring these are being Proposed, from static configuration to dynamic signaling protocols including ANCP. This talk will explore the motivations for MPLS access and then compare the various technologies, both for transport and provisioning.

back to program ^

Network operators are increasingly carrying mission-critical traffic, such as PSTN and mobile voice, broadcast TV and air traffic control radar signals, over IP/MPLS networks. These traffic types tends to be much more "fragile" than the traffic traditionally carried over IP/MPLS networks and so place more stringent requirements on the network equipment in terms of reliability and availability. This talk examines how network equipment has evolved to meet these requirements in order to cater for these types of traffic.

An area that has received great attention over the last couple of years is improving convergence times following link or node failure in IP/MPLS networks. We will therefore discuss in detail convergence mechanisms for both the unicast and multicast over MPLS cases. This discussion will be supported by actual test data and will include a variety of cases including failures in the core and failures at the edge of the network.

Multicast over MPLS schemes are now being used to carry mission-critical Professional Broadcast TV and IPTV traffic. A key requirement for many of these deployments is to have full redundancy of network equipment, including the ingress and egress PEs. In some cases, a “live + live” approach is required where two duplicate traffic flows are sent across the network following diverse paths. In other cases, a “live + standby” approach is required. We will compare the merits of the two schemes and describe how both can be readily achieved using Multicast over MPLS technology. This discussion will be accompanied by actual test results to illustrate the operation of the mechanisms.

This area of work has not been previously discussed in a public forum at this level of detail, so should be of great interest to the MPLS 2008 delegates.

back to program ^

This presentation covers key considerations for Service Providers looking to migrate multicast services to a label switched multicast core network. The different technology options for P-tree building, exchanging customer’s multicast routes, auto discovering peering PEs and encapsulation are described, and the options are compared in terms of factors including ease of migration, how well they address the multicast service requirements and scalability.

mLDP and RSVP-TE extensions for multicast are presented as candidate protocols for P-Tree building and their scaling characteristics are analyzed. PIM and BGP are compared as approaches for exchanging customer’s multicast routes for building Emulated LANs (E-LAN), MI-PMSI trees, Selective-PMSI trees or Partitioned E-LANs.

Possible migration scenarios are presented, highlighting the potential risks associated with different approaches.

back to program ^

Recently there has been a lot of activity in the area of Carrier Ethernet, in particular Provider Backbone Bridges (PBB) and Provider Backbone Bridges – Traffic Engineering (PBB-TE) that provide new data plane capabilities for carriers. In this talk we discuss how architectures based on existing link state technologies (e.g. IS-IS) are finding an application to provide a control plane for Ethernet that can both better utilize network resources and provide carrier grade capabilities. Work ongoing in Provider Link State Bridging (PLSB) as a proposal to IEEE 802.1aq Shortest Path Bridging (SPB) is addressing these new capabilities.

The audience will learn how leveraging a link state protocol simplifies and streamlines the connectivity of PBB and creates per Backbone Edge Bridge (BEB) shortest path trees for the optimal delivery for the entire MEF service set (E-Line, E-LAN, and E-Tree). The proposed control plane leverages link state to replace traditional learning of Ethernet addresses in the Provider network. This enables complete and optimal Multicast and Unicast connectivity. We will discuss carrier aspects such as scalability, OAM capability and stability and loop mitigation techniques.

Key attributes of the Ethernet data plane distinguish the result from that achievable with other technologies. The result is compared and contrasted with the multicast and point to multipoint work that is being done for MPLS and GMPLS and conclusions are drawn.

back to program ^

This presentation discusses the experience gained during the implementation of the packet based WBH service over MPLS PWE. Presented are: the high level design overview, the design requirements for the packet based wireless traffic backhauling; the design of the in-band performance monitoring overlay over the PWE architecture; current deficiencies in standards and implementations, and future requirements. As well, the ideal PWE based WBH network that would satisfy all the requirements for WBH is discussed.

back to program ^

As the majority of the mobile traffic demand shifts to packet based services, Mobile service providers are migrating to IP/MPLS for mobile backhaul (MB). The ability to aggregate Ethernet, TDM, ATM, IP, HDLC and PPP based mobile services over a single transport simplifies the radio access network (RAN) architecture and reduces CAPEX and OPEX. End-to-end resiliency within the RAN presents many challenges. Edge redundancy is required towards the base transceiver station (BTS or Node-B) at the RAN aggregation layer as well as at the RAN-MSC edge. Edge protection must include link and card protection, as well as node protection. Multi-chassis protection using APS (MC-APS) for SONET/SDH can protect stateful ML-PPP, ATM, CES for TDM and other SONET/SDH based services. Similarly, multi-chassis protection using 802.3ad link aggregation groups (MC-LAG) can provide node protection for Ethernet based handoffs (MC-LAG) as well. While MPLS fast-route provides sub 50ms protection within a redundant MPLS core, it can\'t track the multi-chassis based edge level protection across nodes. Pseudowire redundancy is ideal for supporting these MC failover scenarios. Active and standby PWs can be synchronized with the access layer protection and track edge faults. Another consideration is support of a non-stop feature set on the IP/MPLS platforms. Graceful restart is limited in scope and it presents unacceptable scenarios that can cause black-holing of traffic during a transition. The control plane must also protect stateful protocols like ML-PPP; logging and accounting tasks; and Lawful intercept tasks. This presentation will cover the many aspects of deploying highly available IP/MPLS MBH networks.

back to program ^

Mobile traffic is on the rise leading to plenty of challenges for mobile operators in terms of capacity utilization, rising opex and capex. This presentation will explore a few mobile backhaul architectures and evaluate the role of IP/MPLS in evolving mobile backhaul networks towards converged architecture to offer cost effective and resilient solutions to the operators. We will look at some techniques to boost the resiliency and availability of the aggregation part of backhaul (high RAN) networks.

back to program ^

User access technologies such as Wimax are promising to while delivering a number of services including voice, video, data and gaming. These service footprint connect to access and core networks which are themselves in various phases of evolution. For example the access network is being updated to utilize Ethernet services for transport while the MPLS core network is changing to support multiple traffic forms with diverse demands (e.g., scaling, QoS, multicast). This presentation examines the deployment environment of WiMax with respect to the ecosystems in place: mobile networks evolving to 3G/4G, access networks evolving to support triple play and core networks becoming true multi service networks. It presentation leads with a service provider perspective on role of MPLS in backhaul and leads into a discussion of how this impacts the requirements of the service centers in regards to capacity, scalability, QoS, and availability.

back to program ^

This will be available soon.

back to program ^

The increased demand for Ethernet services as specified by the MEF is driving the demand for more efficient and cost effective mechanisms of Ethernet service delivery in the metro. The emerging suite of Ethernet based tools in the form of PBB, PBB-TE and PLSB is exactly the class of technology required.

However frequently there is an existing VPLS network, either in the MAN, in the WAN or both. The incumbent network may have deployed customer facing ports, and/or be underutilized such that there is an expectation that MPLS will lend capacity to the support of Ethernet services. In theory this would be expected to present challenges to service management, OAM and provisioning.

This presentation will provide a brief overview of the IEEE Provider Ethernet toolkit. It will explore the design fundamentals that facilitate Provider Ethernet/VPLS integration, and will explore interworking strategies designed to permit seamless introduction of the provider Ethernet toolkit into existing MPLS/VPLS deployments.

back to program ^

In order to provide SLAs for VPLS services, Service Providers need to ensure its service architecture accommodates a robust service delivery model. A comprehensive approach needs to takes into considerations all the High Availability components.

The presentation defines first the required High Availability components: i.e. Non-Stop-Services, Non-Stop-Routing, In-Service-Upgrade, Service and Network Resiliency.

Then it discusses the service resiliency options and their applicability to VPLS access and to the hand-off between VPLS domains. Multi-Chassis LAG, Active/Standby PWs, MSTP and BGP-based schemes will be considered.

back to program ^

Demand for Ethernet based LAN (E-LAN) services is continuing to rise with customers increasingly asking for E-LAN service at regional and national levels. This service can potentially span across multiple Ethernet metro and MPLS backbone networks.

For an end-to-end service delivery across these networks, numerous issues have to be tackled including MAC address scaling, QoS assurance, service management with OAM, etc. The current design approaches using 802.1ad and H-VPLS have some scalability issues and cannot scale to large number of customers.

In this presentation, we will look at new approaches to create scalable network topologies using VPLS and 802.1ah. We will further discuss ways to provide end-to-end QoS and using 802.1ag as the OAM mechanism for these networks.

back to program ^

MPLS has played a significant role in Qwest's ability to provide robust IP-based services. MPLS in the core has enabled true services convergence. Billions of minutes a month of VoIP traffic co-exists with private Enterprise MPLS VPNs and Internet traffic using separate Layer 2 MPLS security domains to provide traffic separation and traffic prioritization. Starting from Layer 3 MPLS VPNs to Layer 2 MPLS VPNs (e.g.,VPLS) and now IPv6, MPLS technology at the core and edge is enabling an ever expanding set of services to customers.

This talk will expand on the above and touch on:

• Qwest's experience in deploying MPLS-enabled services
• Scalability in the core and the edge
• Providing Quality of Service
• New services introduction enabled by MPLS

back to program ^

This will be available soon.

back to program ^

All main Service Providers are currently working hard on their network infrastructures, previously built to supply each a specific service (telephone, data traffic, leased lines, etc.), in order to reduce costs and boost the introduction of new new multimedia services. These have forced SP, including Telecom Italia, to have a scalable network in terms of capacity and performances being able to face the extraordinary increasing trends shown by the figures.

Telecom Italia has chosen to go to this direction since several yeras when a “unique” IP/MPLS BackBone “OPB” (Optical Packet BackBone-OPB) was built: this is not a slogan on the slides, but the deployment of a life network; not green field but as an evolution of the previous IP backbone.

OPB act as the core on TI NGN network and it a has been so developed from 2001 to aggregate and trans-port all kind of traffic (Fixed Voice, Data and Multi-media) from/to Retail and Wholesale Customer. The optical packet backbone has to assure a very high availability to the Voice traffic (and all other mision critical services), comparable with the PSTN network (the so-called “Five Nines Availability”). That is one of the mission critical aspects and it is achieved by the phisical (total redundant equipments, cards and links) and the logical routing level (MPLS-TE, Fast IGP con-vergence, Preventive Rerouting). The need to encrease performance and scalabilty has adviced TI to insert the new CRS technology in the Core, as all innovation in TI moves as a seismic wave, from the centre to the periphery. Other issues, just like the protection of the multicast traffic or the introduc-tion of CRS Multichassis in the network, are hopefully to be solved in the next future.

The paradigm of a “unique” IP/MPLS BackBone “OPB” reflect itself also in the convergence of services with a particular regards the mobile-fixed services. During the last few years more and more “mobile” cli-ents have increased their demand for ip bandwith or have moved from a traditional TDM transport to an IP trasport (e.g Gn, Gi, MAP signalling).

Also the UTMS/GSM voice (with its signalling) is moving from a TDM core connectivity to an IP trasport. For this project TI has exploit the same solu-tion engineered some years ago for ToiP traffic, a solu-tion based on MPLS-TE.

back to program ^

This presentation covers issues and considerations of the transition from an existing transport network to a next generation packet-based transport network from the viewpoint of not only the enabling networking technology but also the actual deployment and operation. Although the challenge of the transition using such packet-based transport networking technologies as MPLS, PBB-TE or T-MPLS has been discussed so far indicating each technology’s pros and cons, it is still a big concern for telecom operators to choose the appropriate solution. Indeed, existing PSTN, TDM-based leased lines and mobile backhauling services are required to be continuously supported over the packet-based transport network in addition to data-centric services, while still meeting services requirements of guaranteed bandwidth, low latency and jitter, low bit error rate and high resiliency.

Furthermore, imagine that the physical network topology is given, the deployment need to be considered without changing the topology. In this presentation, we describe applicability and assessment of such enabling technologies as MPLS, PBB-TE or T-MPLS to provide both data centric services and existing legacy services and provide the telecom operator’s view on transport technologies. Firstly, those services are characterized and analyzed by the topology of services and the required level of services. For example, the leased line service is generally provided based on the point-to-point-typed service topology and requires high quality and high resiliency with the guaranteed bandwidth. On the other hand, the mobile backhauling service and the Internet access (ISP) service follow hub-and-spoke topology where multiple edge nodes communicate with a center site. The mobile backhauling service requires the same quality of services as the leased line service. The virtual private network (VPN) services are provisioned in the form of multipoint-to-multipoint, and generally requires have only to keep the bandwidth with a statistical multiplexing except access links. Secondly, the applicability of enabling networking technologies to provide these services are compared considering on various physical network topologies such as a mesh or a ring and investigated from the viewpoint of functionalities to achieve high resiliency and guaranteed bandwidth, and the operation. We also describe the issues to continue to provide existing TDM or ATM-based services over such packet based transport network in terms of the delay, the quality of services and the clock synchronization, and describe a direct accommodation of a service in a lower-layer DWDM network for an efficient transport of large capacity services, and interworking operation between the packet-based transport network and the DWDM network for high resiliency.

back to program ^

This presentation discusses a new architecture that has been proposed for the IP Core, called IPoDWDM and integrates DWDM interfaces directly into the router thus eliminating the need for external stand-alone transponders. Furthermore, since resiliency is a key factor in networks and as we all know IP routers have implemented various packet-based protection schemes like IP Fast Reroute (FRR) and MPLS Fast Reroute (FRR) which are known to provide sub-50msec protection times, it is important to highlight that this may not be a satisfactory benchmark for video streaming applications which are the key driving force behind IP traffic growth.

Hence this presentation will cover a new protection scheme for IP networks that takes advantage of the visibility the integrated DWDM interfaces on the routers have into the quality of the optical signal. This in turn allows the routers to realize that the quality of the signal is degrading well before packet errors start to appear, and to move traffic away in a proactive fashion with close to zero packet loss in a much less timeframe than 50msec.

back to program ^

NTT Communications performs a field trial of so call Diamond network build with DWDM and OXC network functions in Tokyo metro area since October 2007.

This presentation provides the result of field tests and the real applications of the GMPLS based optical paths. The test results contain C-band (80 lambda) transmission over DSF and optical path creation by multi-vender GMPLS UNI.

We will provide the specific requirements from view of network provide for network element.

We will also discuss future trial plan and a concept of new services based on the Diamond network.

back to program ^

This presentation describes interoperability effort for ubiquitous GMPLS controlled optical networks, which is called as Kei-han-na interoperability consortium. The consortium focuses on three aspects below, especially aiming at inter-carrier applications;

1. Optical transport network (OTN) interface interoperability
2. Inter-carrier GMPLS network control
3. GMPLS controlled Ethernet.

In the presentation, the detail consortium activity will be presented in terms of standardizing specification proposal, interoperability evaluation and field trial.

back to program ^

The presentation details technologies to cope with challenges when implementing GMPLS in real network and its usage for Japan’s first Layer 1 on demand (L1OD) service.

GMPLS has promising advantages such as quick path provisioning, various protection and restoration, and end-to-end traffic engineering (with path calculation function)), but, the following issues are expected in implementing GMPLS in commercial networks: (i) higher resource utilization (avoiding exclusive use of costly layer 1 resource), (ii) synchronization with OSS to manage, and (iii) high reliability both in data and control planes.

Based on Layer 1 Virtual Private Network frameworks [1][2], we propose a Layer 1 on demand (L1OD) architecture with GMPLS and L1OD server. The server has time based traffic engineering algorithm to optimize and rearrange, if necessary, all future use for higher efficiency [3]. The server receives user demands and triggers semi -permanent GMPLS paths with minimal processing delay at setup for the management mode of the three standardized L1OD models. Gradually, the L1OD server and GMPLS combination can migrate to basic and even enhanced mode. In the basic mode where users can directly request path via GMPLS (UNI mode), the L1OD server plays a traffic engineering server which manages network resources and it also allows co-existence of management mode based L1OD paths. In the enhanced mode where users can use network internal resource information for end-to-end traffic engineering, the combination plays the similar functions and additionally the internal resource virtualization function to let the users know only necessary information. Data plane expects to utilize established protection mechanism such as path protection and VCAT diversity in DTM. Control plane is based on layer 2 or 3 packet technology and thus utilize their reliability techniques, such as BFD and IP routing.

Based on the above techniques, a Japan’s first nation wide operational GMPLS network has started in November 2007. It also launched L1OD service in February 2008. The network, SINET3[4], is a national research and educational network consisting of 75 layer 1 switches with 40Gbps redundant links, servicing more than 700 institutes all over Japan.

The presentation will cover GMPLS operational issues which we found during the network planning and service provisioning of the SINET3 and L1OD.

References: (1) ITU-T Y.1313, “Layer 1 Virtual Private Network service and network architectures,” (2) IETF RFC 4847, “Framework and Requirements for Layer 1 Virtual Private Networks,” (3) Hayashi et al, “Services Parameters and Novel Provisioning Techniques for a Bandwidth Reservation Network,” BoD April 2008, (4) S. Urushidani et al, "Layer-1 Bandwidth on Demand Services in SINET3, " Globecom 2007, Dec. 2007.

back to program ^

The Internet is evolving into a universal medium for exchange of data, information, and knowledge. Over time, most of the information disseminated over the Internet will consist of multimedia content. A Content Delivery Network (CDN) consists of a system of servers and caches that are networked across an IP network and operate cooperatively to deliver multimedia content to end users. This talk will survey developments in content delivery networks and show how they can be deployed in conjunction with MPLS technology to improve network efficiency and enhance end-user service quality.

back to program ^

In today’s fast moving communications industry, service providers need every advantage they can command, particularly as “triple-play and “quad-play” services are rolled out. As and while new technologies emerge constantly, at the core, MPLS is needed to address priorities that respond to consumer needs which are continuously becoming more demanding.

The message for leading service providers is that to thrive, and even to survive, in this competitive environment they must reevaluate the control of the transport in the way complex multi-mode services are delivered to customers. IPTV, both to fixed and mobile targets is at the core of this service “explosion.” In turn, at the core of the service delivery infrastructure is where providers can gain the most in customer service, providing strong gains in market presence, customer loyalty and potentially in revenue. In order to achieve competitive success, services providers need to maintain a delivery environment which can define, augment and modify service and services rapidly. Such an environment must be related to and controlled by business objectives. The control of the network must not driven by physical network limitations or a preexisting more rigid operational support systems infrastructure. To reduce time to market, each service provider must be capable of defining and quickly rolling out support for new service offerings. Service providers also need to be able to offer reduced service activation cycles, self-actuated where possible, to attract and keep new customers.

back to program ^

With nearly 50% of the Internet traffic constituting video and digital traffic growing between 50-100% every year, organizations are optimizing their networks for video delivery. Web conferencing, telepresence, VoDs are various delivery mechanisms that can cut down travel costs and enhance productivity. To address this trend, ISPs and End customers are evaluating bandwidth addition. They are also considering traffic optimization in new ways including service, WAN, application and route optimization. This talk evaluates the requirements for video delivery including HD video over public networks. It also compares and contrasts IP/MPLS and Carrier Ethernet as underlying transport mechanisms for video delivery networks.

back to program ^

Today’s networks face enormous challenges as they evolve to provide “universal services”—any content, anytime, anywhere, on any device. A significant driver of this evolution is the explosion in high definition video generated by users as well as by traditional TV and film distributors.

Service providers must balance their ability to deliver revenue-producing video services with the costs associated with building out their broadband infrastructure. The latency-dependent nature of video transmission poses obvious problems for ensuring quality of experience in a “best effort” network environment. At the same time, bandwidth-intensive peer-to-peer file sharing applications are already morphing to enable broad scale video distribution (eg Joost). In addition, traditional web-based advertising models are evolving from passive instream ads to interactive ad insertion tailored to content delivered to individual users. What technologies can cable providers employ to best deal with these issues?

This session will detail the implications of these video-driven trends on the network, and describe how deep packet inspection (DPI) technology can be used to address these issues. The session will demonstrate how DPI can help cable operators distinguish “good” traffic from “bad”, and use that information to deliver improved quality of service to subscribers.

back to program ^

Associated with the increasing size of MPLS core networks and services is the growing operational challenge to deal with large scale MPLS VPN service provisioning, capacity management, and network troubleshooting.

This presentation provides an overview of the key MPLS operations areas and associated activities to manage MPLS network resources and MPLS VPN services. MPLS operations scenarios are discussed, illustrating how embedded management feature capabilities, such as MPLS OAM and MIBs, can be used to effectively detect and troubleshoot MPLS network failures.

In addition, enhanced feature capabilities are presented, covering enhanced MPLS Traffic Engineering (TE) signaling capabilities (e.g., auto-route/mesh) and enhanced MPLS OAM automation using both embedded automation capabilities as well as application-based management systems.

back to program ^

In this paper, we analyze and compare SONET/SDH OAM standard requirements and Packet OAM standards. Based on this analysis we will discuss OAM features required for the next generation packet transport.

With accelerated shift to carrier grade packet-based networks, the Packet-based OAM are playing an important role. The OAM features for SONET/SDH have truly established its stability, reliability and effectiveness originally developed to support highly reliable transport network for the carriers' services, which is inherited in OTN OAM. However, in terms of the Next Generation Network (NGN), the packet based carrier grade network essentially gives us the new requirements not covered by SONET/SDH OAM framework. This particularly related to the traffic engineering (TE) management, which is not required in SONET/SDH because of its bandwidth guaranteed nature.

The basic categories of OAM are defined as: 1) fault management (FM), 2) performance monitoring (PM) and 3) TE management (TEM). Ethernet OAM ITU-T standard (ETH-OAM/Y.1731) specifies almost equivalent FM feature sets as defined in SONET/SDH standards. The ETH-OAM also defines unique PM features specifically required for the packet transport network, such as packet loss measurement, bandwidth test and delay measurement, etc.

TEM is not defined in either SONET/SDH or ETH-OAM. It is not required in SONET/SDH. The TEM is a mandatory feature when ETH-OEM is applied to a carrier grade packet transport. In this paper, we propose two following TEM features: congestion discovery/notification, and fairness OAM. Congestion discovery/notification is a technique to detect and notify congestion before performance, such as delay and throughput, is defected. It facilitates dynamic hashing-weight change of load balancing and prevents it from performance degradation. Fairness OAM, on the other hand, works to fairly share the resources with the flows. It can also support a fair packet discarding mechanism when the total bandwidth excesses the given bandwidth.

At the conference, we will present details of the new TE management mechanisms mentioned above, which are essential for providing carrier grade packet transport network.

back to program ^

A principal benefit of MPLS is the ability to set up explicitly-routed LSPs for traffic engineering purposes, to better utilize network resources. Traffic engineering approaches commonly involve computing a set of tunnels that optimize global network performance with respect to maximum subscription, delay, utilization, and other performance metrics. In this presentation we will discuss the use of incremental traffic engineering (TE).

Incremental TE represents a tactical approach to traffic engineering that aims to maximize the improvement in network performance while minimizing the number of LSP changes that are required, thereby minimizing associated risk. Incremental TE can occur pro-actively (i.e., as part of an ongoing process) to balance traffic distribution in the network, or re-actively to overcome temporary traffic surges, address local network congestion, or to optimally reroute traffic in the event of a network outage. The solution approach is applicable in a core network exclusively comprising explicitly-routed LSPs, or in a mixed environment of explicitly routed LSPs, dynamically routed LSPs (using CSPF), and LDP-based LSPs.

The presentation will describe the rationale and implementation approaches for incremental TE and present results from a case study.

back to program ^

One of every three MPLS implementations sold by sales cannot be implemented by the provisioning and activation teams. The problem increases exponentially based on the total number of sites, geographic distribution, solution components and delivery partners, not to mention increasingly complex customer requirements. Converting a customer opportunity to a proposal takes weeks or months, only to discover that the “won” deal cannot be implemented due to non-adherence to product rule sets and design best practices.

An additional challenge faced by MPLS product managers pertains to the release of new or updates to existing products and services in response to the dynamic competitive landscape and continuous advances in technology. Communication of these changes among product managers, operational teams and sales teams is often minimal or ineffective. Lack of a common language, well-defined processes, collaborative platforms, standard education and timely information sharing exacerbate the problem. All of these elements make it extremely difficult for sales teams to design, propose and sell solutions successfully based on the most recent product definitions.

In this session we will discuss a unique process that enables peer collaboration around a validated network/service blueprint. A “validated blueprint” based upon the latest MPLS product rules is critical to accurately present a valid solution to your customer. Automating the proposal-creation process with a fully integrated, rules-based system reduces the design rejection rate from one third to close to zero. The ability to share and collaborate around the “validated blueprint” creates a common language among all constituents that ensures the proposal is accurate, leading to a smooth, timely and cost-effective implementation for your customer and creates a baseline reference point for handling any future changes quickly and efficiently. The ability to collect customer and operational requirement and feedback post deployment and modify the validated blueprint as needed, ensures continuous success and customer retention.

back to program ^

This presentation describes the PCE-based backbone network control and management architecture, combining GMPLS, PCE and the management system. Implementation and experiments of this architecture are reported. Finally, end-to-end diverse path computation aspects are discussed, based on state-of-the-art discussion in the IETF.

Traditionally, the management system plays an important role for control and management of backbone networks. GMPLS and PCE are proposed as a way to automate control and management of backbone networks. In the future GMPLS (and MPLS) networks, there is a need to integrate GMPLS, PCE and the management system for (semi-)automatic yet manageable network operations.

The proposed architecture utilizes PCEP (Path Computation Element communication Protocol) between the management system and PCE(s). This architecture has the following two key points.

• The path computation function is separated from the management system to the PCE. This ensures that CPU intensive complicated path computation can be performed in a dedicated server. The management system does not need to keep up-to-date network information. In addition, functional evolution of the management system and the PCE would be easier.
• PCEs are distributed per domain. This ensures scalable path computation, compared to the single PCE case. The PCE needs to contain only per domain network information, not global network information. In addition, computation load can be spread across multiple PCEs.

The feasibility of this architecture is verified through implementation and experiments, where inter-domain LSPs are successfully provisioned.

This presentation further discusses end-to-end diverse path computation aspects. An ability to compute diverse paths is important for improving resiliency. Special attention is needed to compute inter-domain diverse paths, since this is not as straight-forward as in the case of single domain. There are several alternatives being analyzed in the IETF, such as PCE-based simultaneous path computation and sequential path computation. This presentation discusses pros and cons, including performance evaluation, of those alternatives.

back to program ^

Accelerated deployment of dynamic, high-bandwidth applications, such as video, on global MPLS networks drives the need for increased automation with sophisticated decision-support tools. This approach adapts mathematically sophisticated procedures to gather and model detailed data not conveniently available in most networks. The result provides service quality guarantees with minimized expenditures and operational simplicity.

These global networks can be comprised of multiple discrete physical networks and/or multiple overlay networks with independent control schemes. Converging this complicated arrangement of networks involves performing feasibility and implementation analyses. Service differentiation and queue management further drive complexity. These intricate factors coupled with the need for timely results add additional demand on decision-support systems, ultimately testing their usefulness.

In this presentation, we will discuss how some providers have increased automation for specific MPLS network management tasks, including designing and loading the initial LSP mesh, converging disjoint and overlay networks with dissimilar metrics, gathering comprehensive LSP traffic statistics, and managing dynamic LSP bandwidth reservations.

The telecommunications industry is at a critical juncture today. Service providers must re-examine policies and processes to determine how new IP-based services can be delivered in a reliable and profitable manner. Using sophisticated decision-support tools, these providers have accomplished this reliability and have achieved the multiple goals of reducing the cost per Megabit while having scalable network architecture with predictable network performance.

back to program ^