Director of Solution Marketing, IP Division
Alcatel
Despite the availability of a number of metropolitan-area network (MAN) and wide-area network (WAN) data services, many enterprises are still searching for solutions that fully address their evolving WAN requirements. To reduce operating and capital expenditures, many enterprises are considering consolidating their voice, data, and video traffic onto a single converged, packet-based network. Masergy, a U.S.-based supplier of global carrier WAN services, estimates that, on average, cost savings of 30 to 50 percent can be achieved by converging voice, video, data and Internet traffic over a single connection. Convergence allows enterprises to eliminate multiple network connections and reduce costs through greater bandwidth efficiency. In addition, the new structure offers increased support for the latest business applications designed to improve overall business competitiveness and efficiency.
Key components of this architecture are the new generation of carrier Ethernet service platforms and technology solutions, including Ethernet over multiprotocol label switching (MPLS). These innovations enable service providers to offer enterprise customers a higher level of service and reliability than they could get from frame relay (FR) and asynchronous transfer mode (ATM) at a better price. Service providers can leverage the stability and scalability of an MPLS control plane with the flexible bandwidth and economics of Ethernet to deliver competitively priced Layer-2 and Layer-3 WAN services.
Increasingly, a service provider's ability to maintain a competitive market position will depend on its ability to offer WAN Ethernet services. For incumbents, WAN Ethernet services are no longer considered a last-resort defense, but an offensive growth play. Revenues from traditional FR, ATM, and time division multiplex (TDM) private-line services delivered over separate network infrastructures continue to shrink because of competitive pressures and enterprise migration to more flexible and cost-effective Ethernet and advanced IP services.
Enterprises are transitioning from a network-oriented to an application-oriented approach. With the deployment of sophisticated, on-demand applications, traditional point-to-point data services are insufficient. They do not scale economically, become increasingly difficult to manage, and offer limited support for multiple service classes. This is where Ethernet, which is rich in carrier-class features, can excel. However, traditional Ethernet services lack carrier-grade resiliency; scaling; multiple quality of service (QoS) levels; and operations, administration, and maintenance (OA&M) functionality, which limits service scope and operational efficiency.
To move beyond best-effort Ethernet and support enterprise on-demand applications, service providers are expanding their service mix with MPLS-enabled hybrid Layer-2 and Layer-3 WAN Ethernet virtual private networks (VPNs) (see Table 1). These new services will help offset declining legacy data/VPN service revenues. MPLS-based WAN Ethernet VPNs provide the following:
- Scalability, supporting tens of thousands of customers and services without geographical limitations
- Reliability, with sub-50 ms restoration for thousands of simultaneous services
- Hard QoS, to enforce strict quality guarantees on real-time voice, video, and business-critical data applications
- Service provisioning and end-to-end service assurance, for "always-on" services and rapid troubleshooting with Ethernet OA&M
- Service interworking from a FR VPN to an Ethernet virtual leased line (VLL) and support for legacy TDM services
| Point-to-Point Connectivity | Multipoint Connectivity | ||
| Enterprise Profile |
|
|
|
| Service |
VLL (E-Line)
|
IP-VPN
|
VPLS (ELAN)
|
| Enterprise Benefits |
|
|
|
WAN Ethernet scalability requirements have moved beyond cheap, higher user network interface (UNI) bandwidth with the introduction of new, real-time, on-demand enterprise applications. The challenge facing service providers is that they must implement an infrastructure that can scale in many dimensions, including network size and reach and number of customers and services.
MPLS-enabled WAN Ethernet services remove the geographic and size limitations traditionally associated with spanning tree protocol (STP). WAN Ethernet can span beyond metro to national and international networks. Enterprises can connect offices in different countries or continents and appear connected to a single bridged virtual LAN (VLAN). The inherent traffic engineering capabilities of MPLS-enabled WAN Ethernet allows the use of multiple paths to forward traffic. As a result, service providers can use the network to its full potential and avoid or postpone upgrades to expensive high-speed links.
VLAN scalability is greatly improved by overcoming the built-in limitation of 4,096 VLANs through MPLS encapsulation and tunneling. VLANs and stacked VLANs are still used as service delimiters on access links to the enterprise, but they have only local significance. The MPLS-enabled WAN Ethernet network also mitigates the media access control (MAC) address scalability issue because each enterprise customer's MAC addresses only have to be known to the devices at the provider edge. This removes the hard limit on the number of customers, and the service provider network can scale to support tens of thousands of customers and services.
MPLS-based WAN Ethernet enables the service provider to expand its service options for new revenue growth. For example, virtual private LAN service (VPLS) is actively deployed by service providers globally to expand next-generation VPN service offerings beyond IP-VPN. By leveraging IP/MPLS, access diversity and the high-availability capabilities of the network elements, VPLS provides a highly reliable network solution, which allows enterprise customers to seamlessly integrate their local-area networks (LANs) and WANs while ensuring acceptable performance of business-critical applications. VPLS offers the same multipoint benefits as an IP-VPN but at Layer 2. VPLS offers a higher bandwidth choice for Layer-2 multipoint services than FR, with the additional benefit of being protocol-independent. Unlike IP-VPNs, VPLS gives the enterprise complete control over network routing. For some enterprises, maintaining routing control is crucial. VPLS is referred to as ELAN by the Metro Ethernet Forum (MEF).
Ethernet must also scale to support TDM services by emulating T1/E1 or synchronous optical network (SONET)/synchronous digital hierarchy (SDH) circuits. Support for circuit emulation service over Ethernet (CESoE) gives providers a migration path from legacy overlay networks to a converged WAN Ethernet service, allowing them to transparently carry TDM-based voice and data applications, including CES and TDM backhauling. This enables an enterprise to run several services concurrently (VPNs, Internet access, PBX connectivity) over a single physical connection.
Figure 1 illustrates two examples of an MPLS-based Ethernet infrastructure supporting CESoE. TDM traffic flow is integrated, appropriately prioritized, and guaranteed by preserving the full array of framing and signaling protocols so no enhanced TDM features are lost and the enterprise's stringent service level agreement (SLA) is maintained.
Figure 1: Circuit Emulation over a Converged MPLS-Enabled WAN Ethernet
Ensuring QoS is one of the issues facing service providers and a key requirement for enterprises using WAN Ethernet services. The majority of current Ethernet solutions support soft QoS. However, as providers plan to support rapidly growing traffic volumes of both business-critical and delay-sensitive applications, hard QoS becomes necessary as a means of guaranteeing deterministic bandwidth. QoS must be handled in a similar way to connection-oriented TDM and ATM networks.
Soft QoS gives traffic or services relative precedence over other traffic or services in the network. This is mainly achieved by having multiple classes of service (CoSs) in the network with different priority schemes, and by assigning traffic to a specific CoS such as 802.1p and differentiated services (DiffServ).
Soft QoS provides differentiation between traffic types, allowing a service provider to offer multiple service levels, but it fails to account for bandwidth guarantees. With soft QoS, a provider cannot guarantee bandwidth or low packet loss to an enterprise unless the provider significantly overengineers its network to ensure that, in almost all scenarios, traffic is not discarded.
Hard QoS includes mechanisms for guaranteeing the bandwidth for different services and reserving network resources along the path. Hard QoS relies on a connection-oriented approach, where paths are pre-provisioned through the network, and resources allocated along the path, before traffic is forwarded. Plain connectionless technologies such as traditional enterprise Ethernet or pure IP without traffic engineering cannot provide hard QoS.
Figure 2 shows an example of a tailored MPLS-enabled WAN Ethernet service to enterprise customers in a multi-tenant unit. With a full-featured carrier WAN Ethernet solution, the service provider can implement hard QoS to tailor the service to each tenant's requirements. Advanced hierarchical scheduling is implemented through a tiered SLA in which different QoS levels are applied to each service within the WAN VPN. For example, specific committed information rates (CIRs) and peak information rates (PIRs) can be set for each VPLS and application riding the VPN. In addition, an overall QoS level can be assigned to the collection of services used by a customer. This results in the following benefits:
- Service-aware QoS provides optimal service characteristics (delay, loss, jitter, etc.) for performance- and time-sensitive applications to meet the most demanding SLAs.
- Hierarchical scheduling, applied in hardware, can scale to thousands of customers and services without affecting performance.
- Each service or application gets reserved bandwidth, but lower-priority services or applications can burst to use all available bandwidth not required by critical applications. This enables the service provider to avoid over-provisioning bandwidth to guarantee critical application performance.
Hard QoS with hierarchical scheduling offers important WAN Ethernet service attributes for service providers seeking to differentiate their VLL and VPLS service offering from a commoditized, low-margin Ethernet service.
Figure 2: Higher-Value Service Bundling (Voice, Video, Data, CESoE) over a Converged Uplink
The lack of reliability has been a stumbling block to full adoption of Ethernet in the WAN. High availability or "nonstop" performance for WAN Ethernet services is essential to deliver business-critical or lifeline applications. WAN Ethernet must match or improve on the protection capabilities of SONET/SDH for link and node outages.
The increasingly diverse mix of traffic types is rendering spanning tree protocol inadequate for the task. Even with the advent of IEEE standards such as 802.3ad, 802.1w, and 802.1s, which make it possible to combine Ethernet connections for load sharing and better resiliency, reconvergence times can still be excessive depending on the scale of the network. The IETF MPLS fast reroute (FRR) capability enables Ethernet to provide sub-50 ms restoration and is a proven protection mechanism. FRR's inherent scalability enables support for tens of thousands of connections and services, as required in a carrier-class Ethernet infrastructure.
However, as voice, video, and other real-time applications with latency and jitter sensitivities represent more of the traffic load, and require far more stringent SLAs, service providers must aim to improve reliability even further (see Table 2).
| SLA Parameters | Targets by Class of Service | |||
| Measurement | Best Effort (e.g., Internet) |
Priority Business (e.g., CRP/ERP) |
Interactive (e.g., e-learning, e-trading) |
Real-time (e.g., videoconferencing) |
| Network Availability | 99.50% | >=99.999% | >=99.999% | >=99.999% |
| Latency (one way) | n/a | <=15 ms (metro) | <=15 ms (metro) <90 ms (national/global) |
<=15 ms (metro) <90 ms (national/global) |
| Jitter | n/a | n/a | <=1 ms (metro) <5 ms (national/ global) |
<=1 ms (metro) <5 ms (national/global) |
First, service providers must rely on carrier-class Ethernet switches and routers. Product technology enhancements such as nonstop routing and nonstop service augment MPLS FRR to take high availability to an even higher standard.
Figure 3: High Availability is No Longer an Option for WAN Ethernet
Nonstop routing improves on MPLS FRR's sub-50 ms reconvergence through control plane parallelization in the switch/router. As up-to-date routing state information is constantly maintained on the standby control plane module, all topology information and adjacencies are retained during a control plane failure. Nonstop routing provides support for border gateway protocol (BGP), intermediate system to intermediate system (IS-IS) and open shortest path first (OSPF) routing protocols, as well as MPLS signaling protocols. With nonstop routing maintaining the interior and exterior gateway protocols, nonstop service keeps WAN Ethernet services (e.g., VLL, VPLS, IP-VPN) and applications running within the service during a control plane failure. The secondary control module is fully aware of the services and sessions, so there is zero downtime during a control plane switchover. This ensures service level guarantees are maintained at all times, and that WAN Ethernet services can offer rapid and better than SONET-like restoration.
Ethernet OA&M tools are critical for the efficient rollout of WAN Ethernet services. Service providers have been quick to identify this limitation, which has led to increased activity on Ethernet OA&M in the various standards bodies such as IETF (MPLS OA&M), ITU (Y.ethoam), and IEEE (802.1ag Ethernet OA&M and Inter-working between 802.1ag and VPLS), and GMPLS Ethernet label switch workgroup.
MPLS has not only provided the carrier attributes for Ethernet. MPLS-enabled Ethernet provides support for IETF OA&M functionalities, including MPLS tunnel OA&M, pseudowire OA&M, and service level OA&M, to verify continuity, connectivity, and performance. The benefits of MPLS OA&M functions can be realized in the newer generation of carrier Ethernet service platforms, which enable fast service activation and service assurance and deliver guaranteed end-to-end SLAs.
Figure 4 and Table 3 describe the benefits of MPLS OA&M functions, as implemented in the newer generation of carrier Ethernet service platforms. For in-band testing, the OA&M packets closely resemble customer packets to effectively test the customer's forwarding path, but are identifiable so they remain within the service provider's network and are not forwarded to the customer.
Figure 4: Extensive Service-Aware OA&M Toolkit
| Diagnostic Test | Description |
| LSP Ping LSP Traceroute |
|
| SDP Ping | |
| Service Ping | |
| VPLS MAC Ping VPLS MAC Traceroute |
|
| IP VPN Ping IP VPN Traceroute |
|
| Service Mirroring |
A policy-driven approach to OA&M testing is critical in an environment with domain-driven constraints. For example, the customer-located equipment environment may use IEEE ping tests, while the VPLS environment will use MAC or service pings. With service-aware management, a provider can apply service verification policies to a richly modeled WAN Ethernet service, and the service-aware policy will apply the right OA&M test to the right technology on an end-to-end basis.
Ethernet OA&M enables the carrier to provide fast, accurate, and efficient resolution to the customer service problem. Where in the past it may have taken hours or days to resolve an Ethernet issue, it can now be diagnosed in minutes and corrected. The customer SLA is not compromised, and the service provider benefits from reduced operating expenses (OPEX).
Web-based portals allowing customers to monitor the status and health of their WAN Ethernet VPNs can also be enabled with the OA&M functionality. Providers can offer enterprise customers reports showing how their SLA was met over time, and the actual performance experienced by customer traffic over the network. Portals can also enable the enterprise to change QoS levels on one of its services or increase bandwidth when necessary. These types of tools help service providers increase revenues.
Advancements in service-aware management are instrumental to Ethernet's operational savings potential, as confirmed by Ovum-RHK's October 2005 study "VPLS growth requires carrier-class network features." The study reports on eight service providers from Asia, Europe, and North America who provided feedback and data related to their enterprise WAN Ethernet service, specifically VPLS, identified in Figure 5 as A to H. The study identified that the most important operational issues for service providers offering enterprise VPLS services are provisioning, delivery, monitoring, and troubleshooting. The study also identifies the product vendors used to construct the VPLS service.
As a critical operational issue, the study first evaluated the provisioning capabilities of the eight service provider networks. More efficient provisioning capabilities, measured primarily through faster provisioning times, benefit service providers through shorter time to revenue, lower OPEX, and an overall increase in operating asset efficiency. As a second critical operational function, the study evaluated the troubleshooting capabilities and experiences of the eight service providers. Figure 5 details the various times required for VPLS provisioning and troubleshooting.
Figure 5: WAN Ethernet Solution Enables Operational Efficiencies
The Alcatel WAN Ethernet solution was deployed at three of the eight service providers, and enabled 33 to 65 percent operational efficiencies in provisioning and troubleshooting compared to the other vendors. Whether the need is for VPLS, VLL or IP-VPN WAN Ethernet services, the OPEX benefits are real. Major factors contributing to these positive results are operational efficiency improvements and service assurance enabled by Alcatel's Ethernet OA&M toolkit.
Demand from enterprises for networking solutions that enable them to operate efficiently and cost-effectively is strong. They need solutions that can support multiple sophisticated applications, such as supply chain management, enterprise resource planning, customer care, and business-critical operations. Service providers can best meet these requirements by implementing a carrier-class WAN Ethernet solution. This solution will rely on MPLS-enabled WAN Ethernet to provide an architecture that supports the convergence of voice, video, and data services and the delivery of VPN services, including VLL, VPLS, and IP-VPN. With this solution, service providers can tailor service offerings to each enterprise's requirements. When this architecture is built using next-generation carrier Ethernet switches and routers, service providers can count on their network to meet the highest standards in terms of scalability, reliability, availability, and security.
Peter Chahal is Director of Solution Marketing in the IP Division at Alcatel where he is responsible for Managed Communications Services targeting the service provider marketplace with a primary focus on next generation IP/MPLS carrier Ethernet and VPN solutions. Peter has more than 15 years of international experience in product marketing, product management, business development and technical sales in the area of carrier IP and Ethernet solutions and services.
Educational content provided by Alcatel
