Next-Generation Multiservice Networks
Traditional multiservice networks focus on Layer 2 Frame Relay and ATM services, using a common ATM backbone to consolidate traffic. This generation of ATM switches was easily extended to support DSL and cable broadband build-outs.
In contrast, next-generation multiservice networks provide carrier-grade, Layer 3 awareness, such as IP and MPLS, in addition to traditional Layer 2 services. These next-generation multiservice networks can take the form of ATM-, blended IP+ATM-, IP/MPLS-, or SONET/SDH-based networks in order to deliver multiple traffic services over the same physical infrastructure.
Even with the existence of next-generation technology architectures, most providers are not in a position to turn over their core technology in wholesale fashion. Provider technology is often on up-to-decade-long depreciation schedules, and functional life must often parallel this horizon, even if equipment is repurposed and repositioned in the network. Then there is the customer-facing issue of technology service support and migration. Though you might wish to sunset a particular technology, the customer is not often in support of your timetable. This requires a measured technology migration supporting both heritage services along with the latest service features. Next-generation technology versions are often the result, to allow new networking innovations to overlap established network architectures.
The topics of next-generation multiservice switching, Cisco next-generation multiservice ATM switches, and MPLS support on Cisco ATM switches are discussed next.
Next-Generation Multiservice ATM Switching
Next-generation multiservice ATM switching is often defined by a common transmission and switching infrastructure that can natively provide multiple services in such a manner that neither service type interferes with the other. This independence between different services requires a separation of the control and switching planes in multiservice equipment. The control plane acts as the brain, apportioning resources, making routing decisions, and providing signaling, while the switching plane acts as the muscle machine, forwarding data from source to destination.
Separation of the control and switching planes makes it possible to partition the resources of the switching platform to perform multiple services in a native fashion. In much the same way that you can logically partition an IBM mainframe processor into multiple production operating systems, apportioning CPU cycles, memory, storage, and input/output channels to individual logical partitions (LPARs), you can resource partition next-generation multiservice switches to accomplish the same concept of creating multiple logical network services.
Resource portioning in many of the next-generation multiservice switches is accomplished through a virtual switch interface within the control and switching planes. Through a function such as the virtual switch interface, you can have multiple service controllers, each sharing the control plane resources to manage the switching plane, which is the switch fabric that forwards data between a source port and a destination port.
Within the Cisco MGX line of multiservice switches, the virtual switching instance (VSI) allows for an ATM Private Network to Network Interface (PNNI) controller to act as a virtual control plane for ATM services, an MPLS controller to act as a virtual control plane for IP or ATM services, and a Media Gateway Control Protocol (MGCP) controller to act as a virtual control plane for voice services. Each type of controller, through Cisco VSI, directs the assigned resources and interfaces of the physical ATM switch that have been partitioned within its domain of control.
You can run all three controllers and, therefore, multiple services in the same physical ATM switch. If partitioned on a switch, each of these service types is integrated natively and not running as a technology overlay. For example, when running MPLS over an ATM switching fabric, all the network switches run an IP routing protocol and an MPLS label distribution protocol (LDP), which is in contrast to running IP as an overlay via classic ATM permanent virtual circuits (PVCs). Every switch in the MPLS-enabled multiservice network is aware of the multiple services that it provides. The multiple controller capability can allow for a migration from classic ATM switching to MPLS within the same physical architecture.
Figure 3-2 shows the conceptual representation of the Cisco Virtual Switch Architecture. The virtual switch architecture is a Switch Control Interface (SCI) developed by Cisco Systems, Inc., and implemented in the Cisco MGX product line of multiservice switching platforms. The virtual switch works across the control and switching planes, the switching plane essentially performs the traffic-forwarding function. While the control plane and the switching plane represent the workhorse functions of the multiservice switch, within the Cisco design there is also an adaptation plane, a management plane, and an application plane that completes the multiservice system architecture. An example of a requirement for the adaptation plane would be the inclusion of support for Frame Relay services, the adaptation plane facilitating the use of Frame Relay to ATM service interworking. A management plane is required for overall switch control, configuration, and monitoring.
Figure 3-2 Cisco Virtual Switch Architecture
The advantages of next-generation multiservice switching are as follows:
- Multiple service types of ATM, voice, MPLS, and IP are supported on the same physical infrastructure, allowing the provider to leverage both circuit-based and packet-based revenue streams.
- Control plane independence allows you to upgrade or maintain one controller type independently, without interrupting service for other controllers.
- You have the ability to choose and implement a control plane that is best suited to the application requirements.
- The separation of the control and switching planes allow the vendor to develop functional enhancements independently of each other.
- The cost-effective approach of adding MPLS to ATM switch infrastructure allows for the migration to MPLS as a common control plane.
Using next-generation multiservice ATM architectures, providers can maintain existing services such as circuit-based voice and circuit-based video, while migrating to and implementing new packet-based network services such as packet voice, Layer 2 and Layer 3 VPNs, MPLS, and MPLS traffic engineering features. Many providers will maintain ATM infrastructures and might need to bridge from a traditional ATM platform to a next-generation multiservice ATM platform. As an example, Figure 3-3 shows the concept of migrating a Layer 2, full-mesh PVC network to a next-generation multiservice ATM network that uses MPLS rather than discreet PVCs. By adding a Route Processor Module (RPM) to the MGX 8800s in the figure, this next-generation multiservice ATM platform can support Layer 3 IP protocols, and use MPLS to get the best benefits of both routing and switching.
Cisco Next-Generation Multiservice Switches
Using next-generation multiservice network architecture, Cisco offers several solutions that support today's revenue-generating services while accelerating the delivery of new high-value IP-based services. By combining Layer 3 IP and Layer 2 ATM in a straightforward and flexible manner, providers can establish networks that support existing and emerging services. This provides carrier-class data communication solutions that free providers from the economic and technical risks of managing complex multiservice networks.
Cisco implements next-generation multiservice capabilities in the following products:
- Cisco BPX 8600 Series Switches
- Cisco MGX 8250 Series Switches
- Cisco MGX 8800 Series Switches
- Cisco MGX 8900 Series Switches
- Cisco IGX 8400 Series Switches
The next sections describe and compare these Cisco switches.
Cisco BPX 8600 Series Switches
The Cisco BPX 8600 Series Multiservice Switches are IP+ATM platforms providing ATM-based broadband services and integrating Cisco IOS to support MPLS and deliver IP services. The heart of the system is a 19.2 Gbps cross-point switching fabric capable of switching up to two million cells per second, in a multislot chassis. The chassis employs a midplane design, allowing front cards to be adapted to a variety of back cards that provide Layer 1 interface connections such as T3/E3, OC-3/STM-1, and OC-12/STM-4 (622 Mbps). The largest BPX node has a modular, multishelf architecture that scales up to 16,000 DS1s. With heritage from the Cisco acquisition of Stratacom, the BPX switches are often deployed as carrier-class core switches or broadband edge switches in voice, Frame Relay, ATM, wireless, and MPLS provider networks, where OC-12 core links can supply appropriate capacity.
Figure 3-3 Network Migration from Layer 2 to Next-Generation Multiservice ATM Networks
Cisco MGX 8250 Edge Concentrator Switch
The Cisco MGX 8250 Edge Concentrator is a multiservice switch used primarily at the service provider edge supporting narrowband services at 1.2 Gbps of switching capacity. Supporting T1/E1 to OC-12c/STM-4, Ethernet and Fast Ethernet, this switch family is very flexible for providing ATM edge concentration and even MPLS edge concentration where cost-effectiveness is the primary requirement. Switches deployed at the edge of networks need a good balance between port density and cost. The 8250 has 32 card slots for good capacity. A general target for this platform is a maximum capacity of 192 T1s, which would aggregate to 296 Mbps of bandwidth, well under the OC-12/STM-4 uplink capability for this 8250. That leaves bandwidth headroom within the OC-12's 622 Mbps of capacity to also support several Ethernet and a few Fast Ethernet ports. All port cards support hot insert and removal, allowing the provider to add card and port density incrementally in response to demand.
Cisco MGX 8800 Series Switches
The Cisco MGX 8800 Series Multiservice Switches provide significant flexibility at the service provider edge. The Cisco MGX 8800 family is a narrowband aggregation switch with broadband trunking up to OC-48 (2.5 Gbps). The MGX 8800's cross-point switching fabric options operate at either 1.2 Gbps (PXM-1) or up to 45 Gbps (PXM-45) of nonblocking switching. The aforementioned virtual switch architecture allows for multiple control planes via individual controller cards such as PXM-1E for PNNI services, an RPM-PR controller for IP/MPLS services, and a VISM-PR card for packet voice services using MGCP, packet cable Trunking Gateway Control Protocol (TGCP), H.323 video, and Session Initiation Protocol (SIP).
The 8800 series supports narrowband services of T1/E1 ATM, n * T1/E1 inverse multiplexing over ATM (IMA), Frame Relay, high-speed Frame Relay, Systems Network Architecture (SNA), circuit emulation, ATM user network interface (UNI) 3.0/3.1, and switched multimegabit data service (SMDS). These are useful for integrating services such as IP VPNs, Voice over IP (VoIP) and ATM, PPP aggregation, managed intranets, premium Internet services, and IP Fax Relay. Supporting 100 percent redundancy and automatic protection switching (APS), the 8800 series is often deployed as an MPLS multiservice ATM switch on the edges of ATM-based provider networks.
Cisco MGX 8900 Series Switches
The Cisco 8900 Series Multiservice Switch, specifically the 8950, is a high-end multiservice broadband switch designed to scale multiservice networks to OC-192c/STM-64. Supporting a range of broadband services from T3/E3 to OC-192c/STM-64, the MGX 8950 supports the aggregation of broadband services, scaling of MPLS VPNs, and network convergence.
With up to 180 Gbps of redundant switching capacity or 240 Gbps nonredundant, the MGX 8950 is a superdensity broadband switch supporting up to 768 T3/E3s, 576 OC3c/STM-1s, 192 OC-12c/STM-4s, 48 OC-48c/STM-16s, or up to 12 OC-192c/STM-64s in flexible combinations. This switch is specifically architected with a 60 Gbps switch fabric module (XM-60), of which four can be installed to meet the demands and service levels of 10 Gbps ATM-based traffic at the card interface level. The modularity of the XM-60 module allows a provider to incrementally scale switching capacity as needed, starting with one and growing to four per MGX 8950 chassis.
Cisco IGX 8400 Series Switches
Cisco also has a family of multiservice switches that are designed for large enterprises with ATM requirements or for service providers with low cost of ownership requirements. The IGX 8400 series of multiservice WAN switches support line speeds of 64 Kbps up to OC3c/STM-1 with a 1.2 Gbps nonblocking switching fabric. MPLS is also supported on this IP+ATM switch family. The IGX 8400 represents the lowest cost per port of any ATM switch on the market.
Comparing Cisco Next-Generation ATM Multiservice Switches
In summary, the complete family of Cisco multiservice switches support switching speeds from 1.2 Gbps to 240 Gbps, line speeds from DS0 to OC-192c/STM-64 including Fast Ethernet, and ATM edge concentration, PNNI routing, MPLS routing, and packet voice control functions. Both modular and compliant with the various specifications, these products are used to build today's next-generation multiservice ATM networks. Figure 3-4 shows the relative positioning of Cisco next-generation ATM multiservice switches.
Figure 3-4 Cisco Next-Generation ATM Multiservice Switches