Cisco BPX 8600 series wide-area switches offer a variety of service interfaces for data, video, and voice traffic, and support numerous connectivity options to address a broad range of diverse needs. Network interface options include broadband (T3/E3 to OC-12/STM-4) and narrowband (64 kbps to n x T1/E1) via leased lines or public ATM services. Additionally, the BPX switch provides a cost-effective solution by offering a wide range of port densities via the MGX 8220 and MGX 8800 edge concentrators. Proven in the world's largest networks, the Cisco BPX 8620, 8650, and 8680 help you to anticipate and meet market demands while eliminating technology risk.

The Cisco BPX® 8600 Series wide-area switches are standards-based high-capacity broadband ATM switches that provide backbone ATM switching, IP + ATM services including Multiprotocol Label Switching (MPLS) with trunk and CPU hot standby redundancy. The BPX 8600 series deliver a wide range of other user services (see Figure 1-1).

The BPX 8600 Series includes:

BPX 8620 wide-area switch
BPX 8650 IP + ATM switch
BPX 8680 universal service node
BPX 8680-IP (BPX + MGX 8850 + 7204 LSC)

BPX 8620

The Cisco BPX 8620 switch is a scalable, standards-compliant unit, fully compatible with:

Cisco MGX 8800 series wide area edge switch
Cisco MGX 8220 edge concentrator
Cisco IGX 8400 series wide-area switch
Cisco Service Expansion Shelf

The BPX multishelf architecture integrates both IP and ATM services, thereby enabling you to deploy the industry's widest range of value-added services. This architecture offerslow-cost entry points for small sites up to unprecedented port density and scalability for the very largest sites. Finally, it supports both broadband services and narrowband services within a single platform.

The architecture supports both the broadband BPX switch and up to 16 edge concentrator shelves. This scalability results in full utilization of broadband trunks and allows the BPX switch to be expanded incrementally to handle an almost unlimited number of subscribers.

The edge concentrators terminate traffic from a variety of interfaces, such as IP, Frame Relay, ATM, and circuit emulation, and adapt non-ATM traffic into ATM cells. This traffic is aggregated and sent to the BPX switch where it is switched on high-speed ATM links. This aggregation on a single platform maximizes the density of broadband and narrowband ports. High-density aggregation of low-speed services also optimizes the efficiency of the high-speed switching matrix and broadband card slots.

The multishelf view is a "logical" view. Physically, the edge concentrator shelves may be co-located with the BPX switch or they may be located remotely. The connection between a shelf and the BPX switch is a high-speed, optionally redundant ATM link.

The BPX switch consists of the BPX shelf with fifteen card slots that may be co-located with the MGX 8220 or MGX 8800 and Service Expansion Shelf (SES) as required.

Three of the slots on the BPX switch shelf are reserved for common equipment cards. The other twelve are general purpose slots used for network interface cards or service interface cards. The cards are provided in sets, consisting of a front card and its associated back card.

The BPX shelf can be mounted in a rack enclosure that provides mounting for a co-located SES and the MGX 8220 or MGX 8800 interface shelves.

Figure 1-1: BPX Switch General Configuration Example

BPX 8650

The BPX® 8650 is an IP+ATM switch that provides ATM-based broadband services and integrates Cisco IOS® software via Cisco 7200 series routers to deliver Multiprotocol Label Switching (MPLS) services.

The BPX 8650 provides these core Internet requirements:

scalability
advanced IP services
Layer 2 virtual circuit switching advantages
Layer 2/Layer 3 interoperability

The BPX 8650 supports:

Premium IP services
The Internet, intranets, extranets, and IP VPNs, are now available over an ATM infrastructure
Value-added services, such as content hosting, voice over IP, and video, as well as data-managed services
ATM Services
Standards-based ATM interfaces offer broadband and narrowband interconnection for routers, ATM LANs, and other ATM access devices.
The ATM Forum's available bit rate (ABR) virtual source/virtual destination (VS/VD) traffic management capabilities
Constant bit rate (CBR)
Variable bit rate real time (VBR-RT)
VBR nonreal time (VBR-NRT)
Unspecified bit rate (UBR)

BPX 8680

The BPX 8680 universal service switch is a scalable IP+ATM WAN edge switch that combines the benefits of Cisco IOS® IP with the extensive queuing, buffering, scalability, and quality-of-service (QoS) capabilities provided by the BPX 8600 and MGX 8800 series platforms.

The BPX 8680 switch incorporates a modular, multishelf architecture that scales from small sites to very large sites and enables service providers to meet the rapidly growing demand for IP applications while cost-effectively delivering today's services.

The BPX 8680 consists of one or more MGX 8850s connected as feeders to a BPX 8620. Designed for very large installations, the BPX 8680 can scale to 16,000 DS1s by adding up to 16 MGX 8850 concentrator shelves while still being managed as a single node.

BPX 8680-IP

The BPX 8680-IP scalable Layer 2/Layer 3 WAN solution integrating the proven multiservice switching technology of the Cisco BPX 8650 switch with the flexibility and scalability of the Cisco MGX 8850. The MGX 8850 switch serves as an edge concentrator to the BPX 8650, which employs the BPX 8600 series switch modular, multishelf architecture to enable scalability. The BPX 8650 switch includes a Cisco 7204 label switch controller (LSC) and supports multiprotocol label switching (MPLS) for New World integrated infrastructures.

New with Release 9.3

With Release 9.3.0, the BPX switch software supports a number of new features:

Priority Bumping
This feature allows connections for both BPX and IGX that are classified as more important (via COS value) to bump existing connections that are of lesser importance when there are insufficient resources (such as bandwidth) to route these important connections due to trunk failures in the network. You turn on priority bumping, change parameters, and view the statistics by using the command cnfbpparm. This feature cannot be turned on until all nodes are upgraded to 9.3

For procedures on using Priority Bumping, see "Optimizing Traffic Routing and Bandwidth" in the Cisco WAN Switching Command Reference.

UXM ATM Forum IMA Compliant Ports
This feature addresses the need for IMA line support between the IGX and either a router, LS 1010, or an edge device to complete end-to-end interoperability.You can now bundle multiple physical lines into a logical line to enlarge the traffic bandwidth to support high speed ATM without upgrading your access line to higher speed service such as T3/E3 line. By grouping a number of T1/E1 lines with inverse multiplexing of the data flow (ATM Forum IMA protocols) into the group of T1/E1 lines, the group of lines can be treated as a logical high-bandwidth line to solve the narrow bandwidth problem with the advantage of availability and cost-effectiveness.
BXM to BXM-E Upgrades
It is now possible to gracefully "hitlessly" upgrade an active legacy BXM configured in 16K mode to an enhanced BXM-E (DX, EX) configured in 32K mode. You can scale up your networks with the 32K BXM-E on either the port or trunk or a combination of both without any down time and without any service interruption. This feature also supports BXM-E on APS.
Separate Abort Stack
Previously, the BPX and IGX switch software logged both critical and non-critical errors into the Software Error Table. Due to the limited number of entries in the table (12), critical errors (aborts) could be overwritten by non-critical errors, making it hard to determine the cause of faults. The separate Software Abort Table contains only the critical abort faults and retrieved Abort information for reporting and debugging purposes.After an upgrade, old aborts that are stored in the Software Error Table will not be migrated to the new Software Abort Table. Only new aborts will be logged into the Abort table.
Upgrades Protection
This enhancement provides additional protection against running loadrev/runrev and doing upgrades during the time that statistics collection is enabled. This enhancement will warn and automatically disable stats collection if the user says "Yes" to the warning prompt.
VSI MIB Support
Enables the BPX software to track specific information about a VSI controller (such as type, capability, resource usage, and so on). In order for the network amangement system to find out about them, they need to query the controller directly via SNMP. This enhancement is to provide via SNMP MIB the capability to query the BPX switch for VSI controllers attached to that switch and associated information. This allows for easier discovery of BPX-attached VSI controllers by external SNMP-capable application (including Cisco WAN Manager).
Support for <50 cps on connections on the BXM and UXM cards.
With policing turned off this will be supported on all interface types. However, with policing on, the minimum rate will be lowered to 12 to 6 cps only for the T3/E3 and T1/E1 interfaces.
Enhanced Shaping of the Control Traffic
This feature limits the maximum bandwidth guaranteed by the high priority Qbins so that the control traffic does not flood the trunk and overtake the bandwidth allocated for user traffic.
Support for 3 VSI Partitions
The BXM now supports 3 VSI partitions.
800 Board Level Revision Number
The board level revision number (also known as the Manufacturing 800 number) provides the maximum information possible about a given card, which assists in troubleshooting. This enables Cisco Customer Service to remotely identify the board level revision number without physically removing the card from the slot. This project provides the capability to identify the board level revision number via command line interface, Cisco WAN Manager or CiscoView.

Discontinued

These older hardware components and technologies will be supported for five years from the time they are discontinued:

The BNI-155 card
All ASI cards
The BCC-3 card
The BCC-3-32 card
The IPX switch
The Extended Services Processor (ESP)
However, PNNI is available on the BPX via the Service Expansion Shelf (SES) PNNI. For a brief description, see Service Expansion Shelf PNNI.
VSI 1.0
The FastPAD

BPX Switch Operation

With the BCC-4 card, the BPX switch employs a non-blocking crosspoint switch matrix for cell switching that can operate at up to 19.2 Gbps peak. The switch matrix can establish up to 20 million point-to-point connections per second between ports.

The BXM cards support egress at up to 1600 Mbps and ingress at up to 800 Mbps. The enhanced egress rate enhance operations such as multicast.

Access to and from the crosspoint switch matrix on the BCC is through multiport network and user access cards. It is designed to easily meet current requirements with scalability to higher capacity for future growth.

A BPX switch shelf is a self-contained chassis that may be rack-mounted in a standard 19-inch rack or open enclosure.

All control functions, switching matrix, backplane connections, and power supplies are redundant, and non-disruptive diagnostics continuously monitor system operation to detect any system or transmission failure. Hot-standby hardware and alternate routing capability combine to provide maximum system availability.

The BPX Switch with MGX 8220 Shelves

Many network locations have increasing bandwidth requirements due to emerging applications and the confluence of voice, data, and video digital communications. To meet these requirements, you can overlay your existing narrowband networks with a backbone of BPX switches to utilize the high-speed connectivity of the BPX switch operating at up to 19.2 Gbps with its T3/E3/OC-3/OC-12 network and service interfaces.

The BPX switch service interfaces include BXM ports on the BPX switch and service ports on MGX 8220 shelves. The MGX 8220 shelves may be co-located in the same cabinet as the BPX switch, providing economical port concentration for T1/E1 Frame Relay, T1/E1 ATM, CES, and FUNI connections.

Multiprotocol Label Switching

The BPX 8650 MPLS switch combines a BPX switch with a separate MPLS controller (Cisco Series 7200 router). By integrating the switching and routing functions, MPLS combines the reachability, scalability, and flexibility provided by the router function with the traffic engineering optimizing capabilities of the switch.

Multiprotocol Label Switching (MPLS) is a high-performance method for forwarding packets (frames) through a network. It enables routers at the edge of a network to apply simple labels to packets (frames). ATM switches or existing routers in the network core can switch packets according to the labels with minimal lookup overhead.

MPLS integrates the performance and traffic management capabilities of Data Link Layer 2 with the scalability and flexibility of Network Layer 3 routing. It is applicable to networks using any Layer 2 switching, but has particular advantages when applied to ATM networks. It integrates IP routing with ATM switching to offer scalable IP-over-ATM networks.

In contrast to label switching, conventional Layer 3 IP routing is based on the exchange of network reachability information. As a packet traverses the network, each router extracts all the information relevant to forwarding from the Layer 3 header. This information is then used as an index for a routing table lookup to determine the packet's next hop. This is repeated at each router across a network. At each hop in the network, the optimal forwarding of a packet must be again determined.

The information in IP packets, such as IP Precedence information and information on Virtual Private Network membership, is usually not considered when forwarding packets. Thus, to get maximum forwarding performance, typically only the destination address is considered. However, because other fields could be relevant, a complex header analysis must be done at each router that the packet meets.

The main concept of MPLS is to include a label on each packet.

Packets or cells are assigned short, fixed length labels. Switching entities perform table lookups based on these simple labels to determine where data should be forwarded.

The label summarizes essential information about routing the packet:

Destination
Precedence
Virtual Private Network membership
Quality of Service (QoS) information from RSVP
The route for the packet, as chosen by traffic engineering (TE)

With Label Switching the complete analysis of the Layer 3 header is performed only once: at the edge label switch router (LSR) which is located at each edge of the network. At this location, the Layer 3 header is mapped into a fixed length label, called a label.

At each router across the network, only the label need be examined in the incoming cell or packet in order to send the cell or packet on its way across the network. At the other end of the network, an edge LSR swaps the label out for the appropriate header data linked to that label.

A key result of this arrangement is that forwarding decisions based on some or all of these different sources of information can be achieved by means of a single table lookup from a fixed-length label. For this reason, label switching makes it feasible for routers and switches to make forwarding decisions based upon multiple destination addresses.

Label switching integrates switching and routing functions, combining the reachability information provided by the router function, plus the traffic engineering benefits achieved by the optimizing capabilities of switches.

For multiservice networks, the BPX 8650 switch provides ATM, Frame Relay, and IP Internet service all on a single platform in a highly scalable way. Support of all these services on a common platform provides operational cost savings and simplifies provisioning for multiservice providers.

Cisco's MPLS solution is described in detail in the Cisco MPLS Controller Software Configuration Guide.

Private Network to Network Interface (PNNI)

Private Network to Network Interface (PNNI) is a link-state routing protocol that provides standards-based dynamic ATM routing with QoS support as defined by the ATM Forum. PNNI supports aggregation for private ATM addresses and links between switches, and can scale the network and its performance by configuring PNNI peer groups and hierarchical levels.

A key feature of the PNNI hierarchy mechanism is its ability to automatically configure itself in networks in which the address structure reflects the topology. It is responsive to changes in network resources and availability.

PNNI is available on the BPX switch when an optional Cisco Service Expansion Shelf (SES) PNNI is installed. This controller is connected locally to a BPX 8600 series switch to provide PNNI signaling and routing for the establishment of ATM and Frame Relay switched virtual circuits (SVCs) and Soft Permanent Virtual Circuits (SPVCs) over a BPX 8600 wide area network. The network created with BPX SES PNNI nodes also supports traditional ATM and Frame Relay permanent virtual circuits (PVCs) in a separately partitioned AutoRoute network.

ATM SVCs are ATM connections that are established and maintained by a standardized signaling mechanism between ATM CPE (ATM end systems) across a Cisco WAN switching network. ATM SVCs are set up in accordance with user demand and removed when calls are completed, thus freeing up network resources.

BPX SES PNNI node resources, such as port virtual path identifier (VPI) range and bandwidth and trunk bandwidth, are partitioned between SVCs/SVPCs and PVCs. Resource partitioning provides a firewall between PVCs and SVCs/SVPs so that problems with CPE or large bursts do not affect the robustness and availability of PVC services. Bursty data for either PVCs or SVCs/SPVCs can always use any unused link bandwidth, regardless of partitioning.

For a brief description of the SES PNNI, see Service Expansion Shelf PNNI. Refer to the Cisco SES PNNI Controller Software Configuration Guide for detailed information abut the SES.

For further information about PNNI and the SES, refer to the Cisco SES PNNI Controller Software Configuration Guide.

Virtual Private Networks

This section is a brief description of the BPX switch's support for Virtual Private Networks (VPN). For additional information, refer to the Cisco MPLS Controller Software Configuration Guide

Conventional VPNs that use dedicated lease lines or Frame Relay Private Virtual Circuits (PVC) and a meshed network (Figure 1-2) provide many advantages, but typically have been limited in efficiency and flexibility.

Instead of using dedicated leased lines or Frame Relay PVCs, and so on, for a VPN, an IP virtual private network uses the open connectionless architecture of the Internet for transporting data as shown in Figure 1-2.

An IP virtual private network offers these benefits:

Scalability
- Avoids VC mesh configuration
- Easy to add a new site since IP is connectionless
- Service provider handles router service management
Efficient
- Rapid provisioning for networks
- Supports any to any intranets

Figure 1-2: IP VPN Service Example

MPLS Virtual Private Networks

MPLS virtual private networks combine the advantages of IP flexibility and connectionless operation with the QoS and performance features of ATM (Figure 1-3).

The MPLS VPNs provide the same benefits as a plain IP Virtual Network plus:

Scaling and Configuration
- Existing BGP techniques can be used to scale route distribution
- Each edge router needs only the information for the VPNs it supports
- No VPN knowledge in core,
- No need for separate VC mesh per VPN
Highly Scalable
Easy to add new sites
Configure one site on one edge router or switch and network automatically does the rest.
Traffic Separation in MPLS
Each packet has a label identifying the destination VPN and customer site, providing same level of privacy as Frame Relay.
Flexible Service Grouping
Over a single structure can support multiple services, such as voice VPNs, extranets, intranets, Internet, multiple VPNs.

Figure 1-3: MPLS VPNs Example

Frame Relay to ATM Interworking

Interworking lets you retain your existing services and migrate to the higher bandwidth capabilities provided by BPX switch networks, as your needs expand. Frame Relay to ATM Interworking enables Frame Relay traffic to be connected across high-speed ATM trunks using ATM-standard Network and Service Interworking.

Two types of Frame Relay to ATM interworking are supported:

Network Interworking (see Figure 1-4)
- Performed by the BTM card on the IGX switch and
- Performed bythe FRSM card on the MGX 8220
Service Interworking (see Figure 1-5).
- Supported by the FRSM card on the MGX 8220 and
- Supported bythe UFM cards on the IGX switch.

Network Interworking

Part A of Figure 1-4 shows typical Frame Relay to network interworking. In this example, a Frame Relay connection is transported across an ATM network, and the interworking function is performed by both ends of the ATM network.

These are typical configurations:

IGX switch Frame Relay (shelf/feeder) to IGX switch Frame Relay (either routing node or shelf/feeder).
MGX 8220 Frame Relay to MGX 8220 Frame Relay.
MGX 8220 Frame Relay to IGX switch Frame Relay (either routing node or shelf/feeder).

Part B of Figure 1-4 shows a form of network interworking where the interworking function is performed by only one end of the ATM network, and the CPE connected to the other end of the network must itself perform the appropriate service-specific convergence sublayer function.

These are sample configurations:

IGX switch Frame Relay (either routing node or shelf/feeder) to BPX switch or
to MGX 8220 ATM port.
MGX 8220 Frame Relay to BPX switch or MGX 8220 ATM port.

Network Interworking is supported by the FRM, UFM-C, and UFM-U on the IGX switch, and the FRSM on the MGX 8220. The Frame Relay Service Specific Convergence Sublayer (FR-SSCS) of AAL5 is used to provide protocol conversion and mapping.

Figure 1-4: Frame Relay to ATM Network Interworking

Service Interworking

Figure 1-5 shows a typical example of Service Interworking. Service Interworking is supported by the FRSM on the MGX 8220 and the UFM-C and UFM-U on the IGX switch. Translation between the Frame Relay and ATM protocols is performed in accordance with RFC 1490 and RFC 1483.

Unlike Network Interworking, in a Service Interworking connection between an ATM port and a Frame Relay port, the ATM device does not need to be aware that it is connected to an interworking function.

The Frame Relay service user does not implement any ATM specific procedures. Also, the ATM service user does not need to provide any Frame Relay specific functions. All translational (mapping functions) are performed by the intermediate interworking function.

This is a typical configuration for service interworking:

MGX 8220 Frame Relay (FRSM card) to BPX switch or MGX 8220 ATM port.
IGX switch Frame Relay (FRM-U or FRM-C) to BPX switch or MGX 8220 ATM port.

Figure 1-5: Frame Relay to ATM Service Interworking

Tiered Networks

Networks may be configured as:

Flat
All nodes perform routing and communicate fully with one another), or
Tiered
Interface shelves are connected to routing hubs, where the interface shelves are configured as non-routing nodes.

By allowing CPE connections to connect to a non-routing node (interface shelf), a tiered network is able to grow in size beyond that which would be possible with only routing nodes comprising the network.

Starting with Release 8.5, tiered networks support both BPX switch routing hubs and IGX switch routing hubs. Voice and data connections originating and terminating on IGX switch interface shelves (feeders) are routed across the routing network via their associated IGX switch routing hubs.

Tiered networks support multiservice connections, including Frame Relay, circuit data, voice, and ATM. By allowing customer premiswe equipment to connect to a non-routing node (interface shelf), a tiered network is able to grow in size beyond that which would be possible with only routing nodes.

Intermediate routing nodes must be IGX switches. IGX switch interface shelves are the only interface shelves that can be connected to an IGX switch routing hub. With this addition, a tiered network provides a multiservice capability (Frame Relay, circuit data, voice, and ATM).

Routing Hubs and Interface Shelves

In a tiered network, interface shelves at the access layer (edge) of the network are connected to routing nodes via feeder trunks (Figure 1-6).

Routing hubs
Those routing nodes with attached interface shelves are referred to as routing hubs.
Interface shelves
The interface shelves, sometimes referred to as feeders, are non-routing nodes.

The routing hubs route the interface shelf connections across the core layer of the network.The interface shelves do not need to maintain network topology nor connection routing information. This task is left to their routing hubs.

This architecture provides an expanded network consisting of a number of non-routing nodes (interface shelves) at the edge of the network that are connected to the network by their routing hubs.

BPX Switch Routing Hubs

T1/E1 Frame Relay connections originating at IGX switch interface shelves and T1/E1 Frame Relay, T1/E1 ATM, CES, and FUNI connections originating at MGX 8220 interface shelves are routed across the routing network via their associated BPX switch routing hubs.

These requirements apply to BPX switch routing hubs and their associated interface shelves:

Only one feeder trunk is supported between a routing hub and interface shelf.
No direct trunking between interface shelves is supported.
No routing trunk is supported between the routing network and interface shelves.
The feeder trunks between BPX switch hubs and IGX switch interface shelves are either T3 or E3.
The feeder trunks between BPX switch hubs and MGX 8220 interface shelves are T3, E3, or OC-3-c/STM-1.
Frame Relay connection management to an IGX switch interface shelf is provided by Cisco WAN Manager.
Frame Relay and ATM connection management to an MGX 8220 interface shelf is provided by Cisco WAN Manager.
Telnet is supported to an interface shelf; the vt command is not.
Frame Relay connections originating at IGX switch interfaces shelves connected to IGX switch routing hubs may also be routed across BPX switch intermediate nodes.
Remote printing by the interface shelf via a print command from the routing network is not supported.

Figure 1-6: Tiered Network with BPX Switch and IGX Switch Routing Hubs

BPX Routing Hubs in a Tiered Network

Tiered networks with BPX routing hubs have the capability of adding interface shelves/feeders (non-routing nodes) to an IGX/BPX routing network (Figure 1-7). Interface shelves allow the network to support additional connections without adding additional routing nodes.

The MGX 8220 or MGX 8800 and IGX 8400 nodes configured as interface shelves are connected to BPX routing hubs.

The MGX 8220 and MGX 8800 support frame T1/E1, X.21 and HSSI Frame Relay, ATM T1/E1, and CES, and are designed to support additional interfaces in the future.

Figure 1-7: Tiered Network with BPX Routing Hubs

Tiered Network Implementation

These requirements apply to BPX routing hubs and their associated interface shelves:

MGX 8220 Release 4 level is required on all MGX 8220 interface shelves.
Only one feeder trunk is supported between a routing hub and interface shelf.
No direct trunking between interface shelves is supported.
No routing trunk is supported between the routing network and interface shelves.
The feeder trunks between BPX hubs and IGX interface shelves may be T3, E3, or OC-3 (since Release 9.2.30).
The feeder trunks between BPX hubs and MGX 8220 or MGX 8800 interface shelves are T3, E3, or OC-3-c/STM-1.
Frame Relay and ATM connection management to an MGX 8220 or MGX 8800 interface shelf is provide by Cisco WAN Manager
Telnet is supported to an interface shelf; the vt command is not.
Remote printing by the interface shelf via a print command from the routing network is not supported.

Tier Network Definitions

Annex G	Annex G is a bi-directional protocol, defined in Recommendation Q.2931. It is used for monitoring the status of connections across a UNI interface. Tiered Networks use the Annex G protocol to pass connection status information between a Hub Node and attached interface shelf.
BPX Routing Hub	A BPX node in the routing network which has attached interface shelves. Also referred to as a hub node or BPX hub.
MGX 8220 Interface Shelf	A standards-based service interface shelf that connects to a BPX routing hub, aggregrates and concentrates traffic, and performs ATM adapation for transport over broadband ATM networks.
MGX 8800 Interface Shelf	A standards-based service interface shelf that connects to a BPX routing hub, aggregrates and concentrates traffic, and performs ATM adapation for transport over broadband ATM networks.
IGX Interface Shelf	A special configuration of an IGX switch that is connected as a shelf to an IGX routing hub. An IGX interface shelf is sometimes referred to as IGX A/F or feeder. The IGX interface shelf does not perform routing functions nor keep track of network topology.
IGX Routing Hub	An IGX node in the routing network which has attached IGX interface shelves. Also referred to as a hub node or IGX hub.
Feeder Trunk	Refers to a trunk that interconnects an interface shelf with the routing network via a BPX routing hub. A feeder trunk is sometimes referred to as an interface shelf trunk.
IGX/AF	Another name for the IGX interface shelf.
Routing Network	The portion of the tiered network that performs automatic routing between connection endpoints.
VPI	Virtual Path Identifier.
VCI	Virtual Connection Identifier.

Upgrades

Converting an IGX node to an interface shelf requires reconfiguring connections on the node because no upgrade path is provided in changing a routing node to an interface shelf.

A BPX node, acting as a Hub Node, is not restricted from providing any other feature normally available on BPX nodes. A BPX Hub supports up to 16 interface shelves.

Connections within tiered networks consist of distinct segments within each tier. A routing segment traverses the routing network, and an interface shelf segment provides connectivity to the interface shelf end-point. Each of these segments are added, configured and deleted independently of the other segments.

Use the Cisco WAN Manager Connection Manager to configure and control these individual segments as a single end-to-end connection.

Interface shelves are attached to the routing network via a BPX routing hub using a BXM trunk (T3/E3 or OC-3) or BNI trunk (T3/E3). The connection segments within the routing network are terminated on the BNI feeder trunks.

All Frame Relay connection types that can terminate on the BPX are supported on the BNI feeder trunk (VBR, CBR, ABR, and ATF types). No check is made by the routing network to validate whether the connection segment type being added to a BNI feeder trunk is actually supported by the attached interface shelf.

Co-locating Routing Hubs and Interface Shelves
The trunk between an interface shelf and the routing network is a single point of failure, therefore, the interface shelves should be co-located with their associated hub node. Card level redundancy is supported by the Y-Cable redundancy for the BXM, BNI, AIT, and BTM.

Network Management

Communication between CPE devices and the routing network is provided in accordance with Annex G of Recommendation Q.2931. This is a bidirectional protocol for monitoring the status of connections across a UNI interface. (Note: the feeder trunk uses the STI cell format to provide the ForeSight rate controlled congestion management feature.)

Communication includes the real time notification of the addition or deletion of a connection segment and the ability to pass the availability (active state) or unavailability (inactive state) of the connections crossing this interface.

A proprietary extension to the Annex G protocol is implemented which supports the exchange of node information between an interface shelf and the routing network. This information is used to support the IP Relay feature and the Robust Update feature used by network management.

Network Management access to the interface shelves is through the IP Relay mechanism supported by the SNMP and TFTP projects or by direct attachment to the interface shelf. The IP Relay mechanism relays traffic from the routing network to the attached interface shelves. No IP Relay support is provided from the interface shelves into the routing network.

The BPX routing hub is the source of the network clock for its associated feeder nodes. Feeders synchronize their time and date to match their routing hub.

Robust Object and Alarm Updates are sent to a network manager that has subscribed to the Robust Updates feature. Object Updates are generated whenever an interface shelf is added or removed from the hub node and when the interface shelf name or IP Address is modified on the interface shelf. Alarm Updates are generated whenever the alarm state of the interface shelf changes between Unreachable, Major, Minor and OK alarm states.

An interface shelf is displayed as a unique icon in the Cisco WAN Manager topology displays. The colors of the icon and connecting trunks indicate the alarm state of each.

Channel statistics are supported by FRP, FRM, ASI, and MGX 8220 endpoints. BNIs, AITs, and BTMs do not support channel statistics. Trunk Statistics are supported for the feeder trunk and are identical to the existing BNI trunk statistics.

Preferred Routing
Preferred routing within the routing network can be used on all connections. Priority bumping is supported within the routing network, but not in the interface shelves. All other connection features such as conditioning, rrtcon, upcon, dncon, and so on, are also supported.
Local and Remote Loopbacks
Connection local and remote loopbacks are managed at the user interface of the FRP endpoint routing node or interface shelf. Remote loopbacks are not supported for DAX connections. The command addlocrmtlp supports remote loopbacks at FRP DAX endpoints.
Testcon and Testdly
Tstcon is supported at the FRP endpoints in a non-integrated fashion and is limited to a pass/fail loopback test. Fault isolation is not performed. This is the same limitation imposed on inter-domain connections. Intermediate endpoints at the AIT and BNI cards do not support the tstcon feature. Tstdelay is also supported for the FRP and ASI in a non-integrated fashion similar to that of the tstcon command.

Inverse Multiplexing ATM

Where greater bandwidths are not needed, the Inverse Multiplexing ATM (IMA) feature provides a low cost trunk between two BPX switches.

The IMA feature allows BPX switches to be connected to one another over any of the 8 T1 or E1 trunks provided by an IMATM module on an MGX 8220 shelf. A BNI or BXM port on each BPX switch is directly connected to an IMATM module in an MGX 8220 by a T3 or E3 trunk. The IMATM modules are then linked together by any of the 8 T1 or E1 trunks.

Refer to the Cisco MGX 8220 Reference and the Cisco WAN Switching Command Reference publications for further information.

Virtual Trunking

Virtual trunking provides the ability to define multiple trunks within a single physical trunk port interface. Virtual trunking benefits include the following:

Reduced cost by configuring the virtual trunks supplied by the public carrier for as much bandwidth as needed instead of at full T3, E3, or OC-3 bandwidths.
Utilization of the full mesh capability of the public carrier to reduce the number of leased lines needed between nodes in the Cisco WAN switching networks.
Choice of keeping existing leased lines between nodes, but using virtual trunks for backup.
Ability to connect BNI or BXM trunk interfaces to a public network using standard ATM UNI cell format.
Virtual trunking can be provisioned via either a Public ATM Cloud or a Cisco WAN switching ATM cloud.

A virtual trunk may be defined as a "trunk over a public ATM service." The trunk really doesn't exist as a physical line in the network. Rather, an additional level of reference, called a virtual trunk number, is used to differentiate the virtual trunks found within a physical trunk port.

Figure 1-8 shows four Cisco WAN switching networks, each connected to a Public ATM Network via a physical line. The Public ATM Network is shown linking all four of these subnetworks to every other one with a full meshed network of virtual trunks. In this example, each physical line is configured with three virtual trunks.

Figure 1-8: Virtual Trunking Example

Traffic and Congestion Management

The BPX switch provides ATM standard traffic and congestion management per ATM Forum TM 4.0 using BXM cards.

The Traffic Control functions include:

Usage Parameter Control (UPC)
Traffic Shaping
Connection Management Control
Selective Cell Discarding
Explicit Forward Congestion Indication (EFCI)
Priority Bumping

In addition to these standard functions, the BPX switch provides advanced traffic and congestion management features including:

Support for the full range of ATM service types per ATM Forum TM 4.0 by the BXM-T3/E3, BXM-155, and BXM-622 cards on the BPX Service Node.
Advanced CoS Management (formerly Fairshare and Opticlass features) Class of Service management delivers the required QoS to all applications.
- The BPX provides per virtual circuit (VC) queuing and per-VC-scheduling provided by rate controlled servers and multiple class-of-service queuing at network ingress.
- On egress, up to 16 queues with independent service algorithms for each trunk in the network.
Automatic Routing Management (formerly AutoRoute feature), end-to-end connection management that automatically selects the optimum connection path based upon the state of the network and assures fast automatic alternate routing in the event of intermediate trunk or node failures.
Cost-Based Routing Management
ABR Standard with VSVD
Congestion control using RM cells and supported by BXM cards on the BPX Switch.
Optimized Bandwidth Management (formerly ForeSight)
An end-to-end closed loop rate based congestion control algorithm that dynamically adjusts the service rate of VC queues based on network congestion feedback.
Dynamic Buffer Management
Cisco's Frame Relay and ATM service modules are equipped with large buffers and a dynamic buffer management technique for allocating and scaling the buffers on a per VC basis to traffic entering or leaving a node. The switch dynamically assigns buffers to individual virtual circuits based on the amount of traffic present and service level agreements. The large queues readily accommodate large bursts of traffic into the node.
PNNI
A standards-based routing protocol for ATM and Frame Relay SVCs.
Early and partial packet discard for AAL5 connections.

Advanced CoS Management

Advanced Class of Service (CoS) management provides per-VC queueing and per-VC scheduling. CoS management provides fairness between connections and firewalls between connections. Firewalls prevent a single non-compliant connection from affecting the QoS of compliant connections. The non-compliant connection simply overflows its own buffer.

The cells received by a port are not automatically transmitted by that port out to the network trunks at the port access rate. Each VC is assigned its own ingress queue that buffers the connection at the entry to the network. With ABR with VSVD or with Optimized Bandwidth Management (ForeSight), the service rate can be adjusted up and down depending on network congestion.

Network queues buffer the data at the trunk interfaces throughout the network according to the connection's class of service. Service classes are defined by standards-based QoS. Classes can consist of the five service classes defined in the ATM standards as well as multiple sub-classes to each of these classes. Classes can range from constant bit rate services with minimal cell delay variation to variable bit rates with less stringent cell delay.

When cells are received from the network for transmission out a port, egress queues at that port provide additional buffering based on the service class of the connection.

CoS Management provides an effective means of managing the quality of service defined for various types of traffic. It permits network operators to segregate traffic to provide more control over the way that network capacity is divided among users. This is especially important when there are multiple user services on one network. The BPX switch provides separate queues for each traffic class.

Rather than limiting the user to the five broad classes of service defined by the ATM standards committees, CoS management can provide up to 16 classes of service (service subclasses) that you can further define and assign to connections. Some of the COS parameters that may be assigned include:

Minimum bandwidth guarantee per subclass to assure that one type of traffic will not be preempted by another.
Maximum bandwidth ceiling to limit the percentage of the total network bandwidth that any one class can utilize.
Queue depths to limit the delay.
Discard threshold per subclass.

These class of service parameters are based on the standards-based Quality of Service parameters and are software programmable by the user.

Automatic Routing Management

With Automatic Routing Management (formerly referred to as AutoRoute), connections in Cisco WAN switching networks are added if there is sufficient bandwidth across the network and are automatically routed when they are added.

You need enter only the endpoints of the connection at one end of the connection and the IGX switch, and BPX switch software automatically set up a route based on a sophisticated routing algorithm. This feature is called Automatic Routing Management. It is a standard feature on the IGX and BPX switches.

System software automatically sets up the most direct route after considering the network topology and status, the amount of spare bandwidth on each trunk, as well as any routing restrictions entered by the user (for example, avoid satellite links). This avoids having to manually enter a routing table at each node in the network. Automatic Routing Management simplifies adding connections, speeds rerouting around network failures, and provides higher connection reliability.

Cost-Based Routing Management

You can selectively enable cost-based route selection as the route selection per node. With this feature a trunk cost is assigned to each trunk (physical and virtual) in the network. The routing algorithm then chooses the lowest cost route to the destination node. The lowest cost routes are stored in a cache to reduce the computation time for on-demand routing.

Cost-based routing can be enabled or disabled at anytime. There can be a mixture of cost-based and hop-based nodes in a network.

The section, Cost-Based Connection Routing, contains more detailed information about cost-based AutoRoute.

Priority Bumping

Priority bumping allows BPX and IGX switch connections classified as more important (via COS value) to "bump" (that is, set aside) existing connections of lesser importance. While the AutoRoute feature is capable of automatically redirecting all failed connections onto other paths, priority bumping lets you prioritize and sustain more important connections when network resources are diminshed to a point that all connections cannot be sustained. Network resources are reclaimed for the more important connections by bumping (derouting) the traffic on less important connections.

Priority bumping is triggered by insufficient resources (such as bandwidth), resulting from any number events, including changes to the network made by using the commands addcon, upcon, cnfcon, cnnfcos, cnfpref, cnftrk, deltrk. Other triggers include trunk line/card failure, node failure, and comm. failure. The most prominent event is a trunk failure.

For information on setting up Priority Bumping, see "Specifying Priority Bumping" in Chapter 10 of the Cisco WAN Switching Command Reference.

ABR Standard with VSVD Congestion Control

The BPX/IGX switch networks provide a choice of two dynamic rate based congestion control methods, ABR with VSVD and Optimized Bandwidth Management (ForeSight). This section describes Standard ABR with VSVD.

Note ABR with VSVD is an optional feature that must be purchased and enabled on a single node for the entire network.

When an ATM connection is configured between BXM cards for Standard ABR with VSVD per ATM Forum TM 4.0, Resource Management (RM) cells are used to carry congestion control feedback information back to the connection's source from the connection's destination.

The ABR sources periodically interleave RM cells into the data they are transmitting. These RM cells are called forward RM cells because they travel in the same direction as the data. At the destination these cells are turned around and sent back to the source as backward RM cells.

The RM cells contain fields to increase or decrease the rate (the CI and NI fields) or set it at a particular value (the explicit rate ER field). The intervening switches may adjust these fields according to network conditions. When the source receives an RM cell, it must adjust its rate in response to the setting of these fields.

When spare capacity exists with the network, ABR with VSVD permits the extra bandwidth to be allocated to active virtual circuits.

Optimized Bandwidth Management (ForeSight) Congestion Control

The BPX/IGX switch networks provide a choice of two dynamic rate based congestion control methods, ABR with VSVD and Cisco's Optimized Bandwidth Management (ForeSight). This section describes Optimized Bandwidth Management (ForeSight).

Note Optimized Bandwidth Management (ForeSight) is an optional feature that must be purchased and enabled on a single node for the entire network.

Optimized Bandwidth Management (ForeSight) may be used for congestion control across BPX/IGX switches for connections that have one or both end points terminating on cards other than BXM. The ForeSight feature is a dynamic closed-loop, rate-based, congestion management feature that yields bandwidth savings compared to non-ForeSight equipped trunks when transmitting bursty data across cell-based networks.

ForeSight permits users to burst above their committed information rate for extended periods of time when there is unused network bandwidth available. This enables users to maximize the use of network bandwidth while offering superior congestion avoidance by actively monitoring the state of shared trunks carrying Frame Relay traffic within the network.

ForeSight monitors each path in the forward direction to detect any point where congestion may occur and returns the information back to the entry to the network. When spare capacity exists with the network, ForeSight permits the extra bandwidth to be allocated to active virtual circuits. Each PVC is treated fairly by allocating the extra bandwidth based on each PVC's committed bandwidth parameter.

If the network reaches full utilization, ForeSight detects this and quickly acts to reduce the extra bandwidth allocated to the active PVCs. ForeSight reacts quickly to network loading in order to prevent dropped packets. Periodically, each node automatically measures the delay experienced along a Frame Relay PVC. This delay factor is used in calculating the ForeSight algorithm.

With basic Frame Relay service, only a single rate parameter can be specified for each PVC. With ForeSight, the virtual circuit rate can be specified based on a minimum, maximum, and initial transmission rate for more flexibility in defining the Frame Relay circuits.

ForeSight provides effective congestion management for PVC's traversing broadband ATM as well. ForeSight operates at the cell-relay level that lies below the Frame Relay services provided by the IGX switch. With the queue sizes utilized in the BPX switch, the bandwidth savings is approximately the same as experienced with lower speed trunks. When the cost of these lines is considered, the savings offered by ForeSight can be significant.

Network Management

BPX switches provide one high-speed and two low-speed data interfaces for data collection and network management:

High-speed interface
An Ethernet 802.3 LAN interface port is provided for communicating with a Cisco WAN Manager NMS workstation. TCP/IP provides the transport and network layer, Logical Link Control 1 is the protocol across the Ethernet port.
Low-speed interfaces
Two RS-232 ports are provided: one for a network printer and the second for either a modem connection or a connection to an external control terminal. These low-speed interfaces are the same as provided by the IGX switch.

Each BPX switch can be configured to use optional low-speed modems for inward access by the Cisco Technical Response Team for network troubleshooting assistance or to autodial Customer Service to report alarms remotely. If desired, another option is remote monitoring or control of customer premise equipment through a window on the Cisco WAN Manager workstation.

A Cisco WAN Manager NMS workstation connects via the Ethernet to the LAN port on the BPX and provides network management via SNMP. Statistics are collected by Cisco WAN Manager using the TFTP protocol.

You can also use the Cisco WAN Manager's Connection Manager to manage:

Frame Relay connections on IGX switch shelves
Frame Relay and ATM connections on MGX 8220 shelves
MGX 8220 shelf configuration.

Network Management software includes these applications:

Cisco WAN Manager (formerly StrataView Plus)
A single unified management platform utilizing HP OpenView® to manage BPX, IGX, and SES devices.
StrataSphere BILLder
Monitors traffic flow over a network and captures data per standard or customized billing periods and formats.
StrataSphere Modeler
Network modeling tool used for preliminary design of new networks and for analysis and modification studies of existing networks.
StrataSphere Adaptor
Exports network modeling information to external third party modeling systems.
SNMP Service Agent
A service agent that provides an interface for automated provisioning and fault management to customers or Operations Support Systems (OSS).

For further information on network management, refer to the Cisco WAN Manager Operations publication.

Cisco WAN Manager

Cisco WAN Manager is a single unified management platform utilizing HP OpenView® to manage BPX, IGX, and SES devices. It provides a standards-based multiprotocol management architecture. Regardless of the size or configuration of your network, Cisco WAN Manager collects extensive service statistics, tracks resource performance, and provides powerful remote diagnostic and control functions for WAN maintenance.

Online help screens, graphical displays, and easy command line mnemonics make Cisco WAN Manager user-friendly. Plentiful hard disk storage is provided to allow accumulating time of day statistics on many network parameters simultaneously. The data is accumulated by the node's controller card and transmitted to the Cisco WAN Manager workstation where it is stored, processed, and displayed on a large color monitor.

Cisco WAN Manager connects to the network over an Ethernet LAN connection. With Ethernet, you can establish Cisco WAN Manager connectivity to remote nodes via frame relay over TCP/IP to the LAN connector on the local node, or via inband ILMI.

Cisco WAN Manager provides in-band management of network elements via SNMP agent interfaces and MIBs embedded in each node and Interface Shelf. The SNMP agent allows a user to manage a StrataCom network or sub-network from any SNMP-based integrated network management system (INMS).

Connection Management
The Cisco WAN Manager Connection Manager enables you to perform connection provisioning such as adding, configuring, and deleting frame relay, ATM, and frame relay-to-ATM interworking connections.
Network Topology
A map of the network is generated at system installation to graphically display all nodes, trunks, circuit lines, and access devices in the network. Various colors are used to indicate the status of each network item. You can zoom in to display specific network details while a small overview map remains displayed as a locator.
Network Performance
Statistics are collected and temporarily stored by each node in the network and released to Cisco WAN Manager when you enable polling, and in accordance with your configuration for specific information within reports. Cisco WAN Manager then stores statistics in a relational database; you retrieve and view these statistics by invoking a statistics display window from the Cisco WAN Manager GUI. From data gathered throughout the network, you can quickly view the operational integrity and deployment of installed network devices and communication media by activating and invoking statistics displays.
Equipment Management
The Cisco WAN Manager Equipment Manager provides the ability to perform equipment management functions such as adding lines and ports on a Cisco MGX 8220 edge concentrator shelf.
Alarm Reporting/Event Log
Cisco WAN Manager displays major and minor alarm status on its topology screen for all nodes in a network. It also provides an event log with configurable filtering of the log events by node name, start time, end time, alarm type, and user specified search string.
Software Updates
System software and software updates are supplied on magnetic tape or floppy disk. You can then load the system software files onto the Cisco WAN Manager workstation where they can be downloaded to a buffer memory in each node in the network in a background mode without disturbing network operation. When the loading is complete for all nodes, you issue a command to switch all nodes over to the new software. The previous software is preserved and can be recalled at any time.
Backup
You can obtain all network configuration files from the network and store them on the Cisco WAN Manager workstation for backup purposes. In the event of a system update or a node failure, you can download the configuration files to one or all nodes for immediate system restoration.

Network Interfaces

Network interfaces connect the BPX switch to other BPX or IGX switches to form a wide-area network. The BPX switch provides these trunk interfaces:

T3
E3
OC-3/STM-1
OC-12/STM-4

The T3 physical interface utilizes DS3 C-bit parity and the 53-byte ATM physical layer cell relay transmission using the Physical Layer Convergence Protocol.

The E3 physical interface uses G.804 for cell delineation and HDB3 line coding.

The BXM-622 cards support these physical interfaces:

SMF
SMFLR

The BPX switch supports network interfaces up to 622 Mbps and provides the architecture to support higher broadband network interfaces as the need arises.

Optional redundancy is on a one-to-one basis. The physical interface can operate either in a normal or looped clock mode. As an option, the node synchronization can be obtained from the DS3 extracted clock for any selected network trunk.

Service Interfaces

Service interfaces connect ATM customer equipment to the BPX switch. ATM User-to-Network Interfaces (UNI) and ATM Network-to-Network Interfaces (NNI) terminate on the ATM Service Interface (ASI) cards and on BXM T3/E3, OC-3, and OC-12 cards configured for as service interfaces (UNI access mode).

The BXM T3/E3 card supports the standard T3/E3 interfaces.

The BXM-155 cards support SMF, SMFLR, and MMF physical interfaces.

The BXM-622 cards support SMF and SMFLR physical interfaces.

The BXM cards support cell relay connections that are compliant with both the physical layer and ATM layer standards.

The MGX 8220 interfaces to a BNI or BXM card on the BPX, via a T3, E3, or OC-3 interface. The MGX 8220 provides a concentrator for T1 or E1 Frame Relay and ATM connections to the BPX switch with the ability to apply Optimized Bandwidth Management (ForeSight) across a connection from end-to-end. The MGX 8220 also supports CES and FUNI (Frame Based UNI over ATM) connections.

Statistical Alarms and Network Statistics

The BPX Switch system manager can configure alarm thresholds for all statistical type error conditions. Thresholds are configurable for conditions such as frame errors, out of frame, bipolar errors, dropped cells, and cell header errors. When an alarm threshold is exceeded, the NMS screen displays an alarm message.

Graphical displays of collected statistics information, a feature of the Cisco WAN Manager NMS, are a useful tool for monitoring network usage. Statistics collected on network operation fall into four general categories:

Node statistics
Network trunk statistics
Network Service, line statistics
Network Service, port statistics

These statistics are collected in real-time throughout the network and forwarded to the WAN Manager workstation for logging and display. The link from the node to the Cisco WAN Manager workstation uses a protocol to acknowledge receipt of each statistics data packet.

Refer to the Cisco WAN Manager Operations publication, for more details on statistics and statistical alarms.

Node Synchronization

A BPX Service switch network provides network-wide, intelligent clock synchronization. It uses a fault-tolerant network synchronization architecture recommended for Integrated Services Digital Network (ISDN). The BPX switch internal clock operates as a Stratum 3 clock per ANSI T1.101.

Because the BPX switch is designed to be part of a larger communications network, it is capable of synchronizing to higher-level network clocks as well as providing synchronization to lower-level devices. You can configure any network access input to synchronize the node. Any external T1 or E1 input can also be configured to synchronize network timing.

A clock output allows synchronizing an adjacent IGX switch or other network device to the BPX switch and the network. In nodes equipped with optional redundancy, the standby hardware is locked to the active hardware to minimize system disruption during system switchovers.

You can configure the BPX Service Node to select clock from these sources:

External (T1/E1)
Line (DS3/E3)
Internal

Switch Software Description

The Cisco WAN switching cell relay system software shares most core system software, as well as a library of applications, between platforms. System software provides basic management and control capabilities to each node.

BPX node system software manages its own configuration, fault-isolation, failure recovery, and other resources. Because no remote resources are involved, this ensures rapid response to local problems. This distributed network control, rather than centralized control, provides increased reliability.

Software among multiple nodes cooperates to perform network-wide functions such as trunk and connection management. This multiprocessor approach ensures rapid response with no single point of failure. System software applications provide advanced features that you can install and configure as required.

Some of the many software features are:

Automatic routing of connections (Automatic Routing Management feature).
Various classes of service that may be assigned to each connection type (Advanced CoS Management).
Bandwidth reservation on a time-of-day basis.
Detection and control of network congestion with ABR with VSVD or Optimized Bandwidth Management (ForeSight) algorithms.
Automatic self-testing of each component of the node.
Automatic collecting and reporting of many network-wide statistics, such as trunk loading, connection usage, and trunk error rates, as you specify.

The system software, configuration database, and the firmware that controls the operation of each card type is resident in programmable memory and can be stored off-line in the Cisco WAN Manager NMS for immediate backup if necessary. This software and firmware is easily updated remotely from a central site or from Customer Service, which reduces the likelihood of early obsolescence.

Connections and Connection Routing

The routing software supports the establishment, removal and rerouting of end-to-end channel connections. There are three routing modes:

Automatic Routing
The system software computes the best route for a connection.
Manual Routing
You can specify the route for a connection.
Alternate Routing
The system software automatically reroutes a failed connection.

The system software uses these criteria when it establishes an automatic route for a connection:

Selects the most direct route between two nodes.
Selects unloaded lines that can handle the increased traffic of additional connections.
Takes into consideration user-configured connection restrictions (for example whether or not the connection is restricted to terrestrial lines or can include satellite hops or routes configured for route diversity).

When a node reroutes a connection, it uses these criteria and also looks at the priority that has been assigned and any user-configured routing restrictions. The node analyzes trunk loading to determine the number of cells or packets the network can successfully deliver. Within these loading limits, the node can calculate the maximum combination allowed on a network trunk of each type of connection: synchronous data, ATM traffic, Frame Relay data, multimedia data, voice, and compressed voice.

Network-wide T3, E3, OC-3, or OC-12 connections are supported between BPX switches terminating ATM user devices on the BPX switch UNI ports. These connections are routed using the virtual path and/or virtual circuit addressing fields in the ATM cell header.

Narrowband connections can be routed over high-speed ATM backbone networks built on BPX broadband switches. FastPacket addresses are translated into ATM cell addresses that are then used to route the connections between BPX switches, and to ATM networks with mixed vendor ATM switches. Routing algorithms select broadband links only, avoiding narrowband nodes that could create a choke point.

Connection Routing Groups

The re-routing mechanism ensures that connections are presorted in order of cell loading when they are added. Each routing group contains connections with loading in a particular range. The group containing the connections with the largest cell loadings is rerouted first, and subsequent groups are then rerouted on down to the last group that contains connections with the smallest cell loadings.

There are three configurable parameters for configuring the rerouting groups:

Total number of rerouting groups
Starting load size of first group
Load size range of each group

You configure the three routing group parameters by using the cnfcmparm command.

For example, there might be 10 groups, with the starting load size of the first group at 50, and the incremental load size of each succeeding group being 10 cells. Then group 0 would contain all connections requiring 0-59 cell load units, group 1 would contain all connections requiring from 60-69 cell load units, on up through group 9 which would contain all connections requiring 140 or more cell load units.

Table 1-1: Routing Group Configuration Example

Routing group	Connection cell loading

0	0-59
1	60-69
2	70-79
3	80-89
4	90-99
5	101-109
6	110-119
7	120-129
8	130-139
9	140 and up

Cost-Based Connection Routing

In standard AutoRoute, the path with the fewest number of hops to the destination node is chosen as the best route. Cost-based route selection uses an administrative trunk cost routing metric. The path with the lowest total trunk cost is chosen as the best route.

Cost-based route selection is based on Dijkstra's Shortest Path Algorithm, which is widely used in network routing environments. You can use cost-based route selection (that is, cost-based AutoRoute) to give preference to slower privately owned trunks over faster public trunks that charge based on usage time. This gives network operators more control over the usability of their network trunks, while providing a more standard algorithm for route selection.

Major Features of Cost-Based AutoRoute

Here is a short description of the major functional elements of Cost-Based Route Selection.

Enabling Cost-Based Route Selection.
You enable cost-based route selection at any time. This feature does not require special password access. The default algorithm is the hop-based algorithm.
Configuring Trunk Cost
You assign a trunk cost to each trunk (physical and virtual) in the network. One cost is assigned per trunk; no separate costs are used for different connection or service types. The valid range of trunk costs is 1 (lowest cost) to 50 (highest cost). A trunk has a default cost of 10 upon activation. The cost of a trunk can be changed before or after the trunk has been added to the network topology.

The cost can also be changed after connections have been routed over the trunk. Such a change does not initiate automatic connection rerouting, nor does it cause any outage to the routed connections. If the new trunk cost causes the allowable route cost for any connections to be exceeded, the connections must be manually rerouted to avoid the trunk. This avoids large-scale simultaneous network-wide rerouting and gives you control over the connection reroute outage.

Cache vs. On-Demand Routing
In previous releases, Hop-Based Route Selection always requires on-demand routing. On-demand routing initiates an end-to-end route search for every connection. Due to the computation time required for Dijkstra's algorithm in cost-based route selection, a route cache is used to reduce the need for on-demand routing.

This cache contains lowest cost routes as they are selected. Subsequent routing cycles use these existing routes if the routing criteria are met. Otherwise on-demand routing is initiated. This caching greatly benefits environments where routing criteria is very similar among connections.

Enabling cost-based route selection automatically enables cache usage. Enabling Hop-Based Route Selection automatically disables cache usage. Cache usage can also be independently enabled or disabled for both types of route selection.

On-Demand Lowest Cost Route Determination
On-demand routing chooses the current lowest cost route to the destination node. This lowest cost route is bounded by the maximum route length of 10 hops. If more than one route of similar cost and distance is available, the route with most available resources is chosen. No route grooming occurs after the initial routing. A connection does not automatically reroute if its route cost changes over time. A connection also does not automatically reroute if a lower cost route becomes available after the initial routing. However, a forced reroute or a preferred route can be used to move the connection to a lower cost route.
Delay-Sensitive Routes
Delay-sensitive IGX connection types (Voice and Non-Timestamped Data) may be configured to use the worst case queueing delay per trunk, rather than the configured trunk cost, in the lowest-cost route determination. The trunk delay acts as the cost attribute in the Dijkstra algorithm. The default mode for the delay sensitive connections is to use the trunk cost. All other connection types always use the trunk cost in the route determination.

AutoRoute does not use the worst case end-to-end queueing delay in route selection for delay sensitive BPX connection types (ATM CBR). Cost-based route selection does not change this.

Cost Cap
A maximum allowable cost value (cost cap) is used during route determination to prevent selection of a route which exceeds an acceptable cost. For routing based on delay, the cost cap is the acceptable end-to-end delay for the connection type. This cap is configured network-wide per delay sensitive connection type.

For routing based on trunk cost, the cost cap is the acceptable end-to-end cost. This cap is configured per connection. The default cost cap is 100, which is derived from the maximum hops per route (10) and default cost per trunk (10). You can change the cost cap at any time. If the cost cap is decreased below the current route cost, the connection is not automatically rerouted. A manual reroute is required to route the connection to fit under the new cost cap. This gives you more control over the connection reroute outage.

Hop-Based Route Selection
Since Release 9.0, AutoRoute uses Hop-Based Route Selection. The cost of all trunks is set to the default cost (10). The cost cap of all connections is set to the maximum allowable cost (100). All other new cost-based routing parameters are set to regular default values.
AutoRoute Interoperability
Because AutoRoute is source-based, nodes can interoperate using different route selection algorithms. The originating node computes the full end-to-end route based on its own knowledge of the network topology. The route is then passed to the subsequent nodes on the route. This source routing allows a mix of Cost-Based and Hop-Based Route Selection to run in a network.

Cost-Based AutoRoute Commands

You use these switch software Command Line Interface (CLI) commands for cost-based route selection:

cnfcmparm
Enables cost-based route selection. This is a SuperUser command to configure all AutoRoute parameters. By default cost-based route selection is disabled. Enabling or disabling cost-based route selection can be done at any time. Each connection routing cycle uses whichever algorithm is enabled when the cycle begins. The configuration is node-based, not network-based, which allows each node to have its own route selection algorithm.

Enabling cost-based route selection automatically enables cache usage. Disabling cost-based route selection automatically disables cache usage. Cache usage may also be independently enabled or disabled.

cnftrk
Configures the administrative cost for a trunk. Both physical and virtual trunks have the cost attribute. Each trunk has a cost ranging from 1 (lowest) to 50 (highest). The default cost is 10 upon trunk activation.

The cost can be configured from either end of the trunk. The cost can be changed before or after the trunk has been added to the network. The cost can also be changed after connections have been routed over the trunk. Any cost change is updated network-wide. Every node in the network stores the cost of every trunk in the network. This knowledge is required for successful source-based routing.

cnfrtcost
Configures the cost cap for a connection. This command is valid only at the node where the connection is added.
cnfsysparm
Configures the delay cost cap for all delay sensitive connections in the network.
dspcon
Displays the maximum and current costs for a connection route
dspload
Displays the administrative cost and queue delay for a network trunk
dsprts
Displays the current costs for all connection routes
dsptrkcnf
Displays the configured cost of a trunk

The Cisco WAN Switching Command Reference contains detailed information about the use of BPX switch commands.

Network Synchronization

Cisco WAN switching cell relay networks use a fault-tolerant network synchronization method of the type recommended for Integrated Services Digital Network (ISDN). You can select any circuit line, trunk, or an external clock input to provide a primary network clock. Any line can be configured as a secondary clock source in the event that the primary clock source fails.

All nodes are equipped with a redundant, high-stability internal oscillator that meets Stratum 3 (BPX) or Stratum 4 requirements. Each node keeps a map of the network's clocking hierarchy. The network clock source is automatically switched in the event of failure of a clock source.

There is less likelihood of a loss of data resulting from re-frames that occur during a clock switchover or other momentary disruption of network clocking with cell-based networks than there is with traditional TDM networks. Data is held in buffers and packets are not sent until a trunk has regained frame synchronism to prevent loss of data.

Switch Availability

Cisco WAN hardware and software components are designed to provide a switch availability in excess of 99.99%. Network availability will be impacted by link failure, which has a higher probability of occurrence than equipment failure.

Because of this, Cisco WAN network switches are designed so that connections are automatically rerouted around network trunk failures, often before users detect a problem. System faults are detected and corrective action taken often before they become service affecting. This section describes some of the features that contribute to network availability.

Node Redundancy

System availability is a primary requirement with the BPX switch. The designed availability factor of a BPX switch is (99.99%) based on a node equipped with optional redundancy and a network designed with alternate routing available. The system software, as well as firmware for each individual system module, incorporates various diagnostic and self-test routines to monitor the node for proper operation and availability of backup hardware.

For protection against hardware failure, a BPX switch shelf can be equipped with the following redundancy options:

Redundant common control modules
Redundant crosspoint switch matrixes
Redundant high-speed data and control lines
Redundant power supplies
Redundant high-speed network interface cards
Redundant service interface cards

If redundancy is provided for a BPX switch, when a hardware failure occurs, a hot-standby module is automatically switched into service, replacing the failed module. All cards are hot-pluggable, so replacing a failed card in a redundant system can be performed without disrupting service.

Since the power supplies share the power load, redundant supplies are not idle. All power supplies are active; if one fails, then the others pick up its load. The power supply subsystem is sized so that if any one supply fails, the node will continue to be supplied with adequate power to maintain normal operation of the node. The node monitors each power supply voltage output and measures cabinet temperature to be displayed on the NMS terminal or other system terminal.

Node Alarms

Each BPX switch shelf within the network runs continuous background diagnostics to verify the proper operation of all active and standby cards, backplane control, data, and clock lines, cabinet temperature, and power supplies. These background tests are transparent to normal network operation.

Each card in the node has front-panel LEDs to indicate active, failed, or standby status.

Each power supply has green LEDs to indicate proper voltage input and output.

An Alarm, Status, and Monitor card collects all the node hardware status conditions and reports it using front panel LED indicators and alarm closures. Indicators are provided for major alarm, minor alarm, ACO, power supply status, and alarm history. Alarm relay contact closures for major and minor alarms are available from each node through a 15-pin D-type connector for forwarding to a site alarm system.

BPX switches are completely compatible with the network status and alarm display provided by the Cisco WAN Manager NMS workstation. In addition to providing network management capabilities, it displays major and minor alarm status on its topology screen for all nodes in a network.

The Cisco WAN Manager NMS also provides a maintenance log capability with configurable filtering of the maintenance log output by node name, start time, end time, alarm type, and user-specified search string.

Table of Contents

The BPX Switch: Functional Overview