PNNI Network Planning Guide for MGX and SES, Release 2.1
Interoperability and Performance Planning

Table of Contents

Interoperability and Performance Planning
Compatible Standards
Specifications
Physical Network Planning
Planning Guidelines for Individual Peer Groups
Planning Guidelines for Hierarchical Networks

Interoperability and Performance Planning


This chapter describes the standards supported by the Cisco switches covered in this guide and provides performance specifications for these switches.

Compatible Standards

The Cisco switches described in this guide are designed to interoperate with switches that support the following standards:

  • PNNI Version 1
  • UNI 3.0
  • UNI 3.1
  • UNI 4.0
  • ILMI 4.0
  • AINI
  • IISP
  • TM

Specifications

Table 2-1 lists PNNI network specifications for the PNNI switches covered in this guide.

Table 2-1   PNNI Networking Specifications

Capabilities BPX/SES 1.0 BPX/SES 1.1 MGX 8850 2.0 MGX 8850 2.1 MGX 8950 2.1

Connections (SVCs and SPVCs) per switch

50 K

100 K

50 K

100 K

100 K

Connections (SVCs and SPVCs) per card

32 K

60 K

32 K

60 K

60 K

Connections (SVCs and SPVCs) per port

32 K

32 K

32 K

32 K

32 K

Physical ports per switch

100

100

100

100

100

PNNI links per switch

100

100

100

100

100

UNI links per switch

100

100

100

100

100

PNNI nodes per peer group

256

190

256

160

160

PNNI peer groups per network

no limit

189

no limit

159

159

Hierarchical levels per PNNI network

1

10

1

10

10

ATM addresses per port

256

256

256

256

256

ATM addresses per switch

3000

2000

3000

2000

2000

Calls per second

100

100

100

100 (PXM45)
250 (PXM45B)

250 (PXM45B)

SPVC reroutes per second (based on service class)

>80

>80

>125

>125

>125

Call latency

<10 ms

<10 ms

<10 ms

<10 ms

<10 ms

Convergence time for network updates

<20 secs

<20 secs

<20 secs

<20 secs

<20 secs

Link failure detection minimum period

12.5 ms

12.5 ms

12.5 ms

12.5 ms

12.5 ms

Link failure detection maximum period

42.5 ms

42.5 ms

42.5 ms

42.5 ms

42.5 ms

Physical Network Planning

The PNNI switches described in this guide can reroute connections and adjust to equipment or link failures only when the physical network has been designed to use redundant hardware and links. When designing a PNNI network, consider the following:

  • Install redundant hardware in switches
  • Install parallel links between adjacent switches
  • Set up multiple links between adjacent peer groups
  • Use multiple links when connecting to an external network
  • Provide multiple communication paths between any two nodes that will communicate with each other

The following sections provide additional information on these guidelines.

Install Redundant Hardware in Switches

The switches described in this guide support redundant power supplies, Processor Switch Module (PXM) cards, line cards, and trunk cards. Although PNNI can reroute calls, using redundant hardware can improve network stability and performance by preventing reroutes caused by hardware failure.

Parallel Links Between Adjacent Switches

When parallel links are available between a pair of switches, PNNI can load balance calls across these links. If one link fails, the other is still available.

Multiple Links Between Adjacent Peer Groups

Communications between two peer groups takes place through two border nodes. Parallel links between two border nodes improves reliability. Adding additional border nodes to handle communications between two peer groups provides alternative routing paths and prevents network outages caused by a single node failure.

Multiple Links to an External Network

An external network connection is any non-PNNI network connection. External network connections include AINI and IISP connections. As with internal network links, consider using parallel links and additional border nodes to provide alternative paths to external networks. When you configure multiple static links to an external network, remember to duplicate the ATM address advertisement configuration on all redundant links.

Multiple Paths Between Network Nodes

It is good design practice to ensure that there are at least two different paths between any pair of nodes that will communicate with each other. A pair of redundant links is one path. If one switch site is damaged by fire or earthquake, there should be at least one other switch that can provide an alternative path between the source and destination switches.

Planning Guidelines for Individual Peer Groups

The first step in planning a PNNI topology is to determine if all network nodes will participate in one peer group or in hierarchical peer groups. The single and hierarchical peer group topologies are introduced in "Introduction to PNNI." When a network grows beyond the capabilities of a single peer group, you must use the hierarchical peer group topology.

The planning for a single peer group topology is the same as the planning for a single peer group within a hierarchical PNNI network. The difference between planning for a single peer group network and planning for a hierarchical network is that for hierarchical networks, you have to plan the communications between peer groups. The following list summarizes the capabilities and guidelines for a single peer groups:

  • For networks with less than 100 nodes, Cisco Systems recommends using a single peer group topology.
  • A single peer group supports up to 256 nodes as described in Table 2-1.
  • All nodes within a single peer group must be adjacent to each other. If the only path between nodes A and B in a peer group is through another peer group, PNNI advertisements between nodes A and B will be blocked at the borders of the other peer group.
  • PNNI can route calls through a maximum of 20 nodes within a single peer group. The structure of the peer group should be such that no node is more than 20 hops from any other node.
  • As the number of nodes in a single peer group grows, the network and system resource requirements for each node grows.
  • The level of PNNI processing required for your network is dependent on the service classes supported. For example, CBR and VBR calls consume more resources than ABR and UBR calls.
  • Although the number of network nodes in your network might not dictate a hierarchical topology, other network requirements can. For example, if you want to take advantage of the individual peer group management capabilities in a split-border hierarchy, you must configure multiple peer groups.
  • Consider future growth when planning peer groups. If the number of existing nodes is approaching the limit of a single peer group, consider using a hierarchical topology so that you do not have to reconfigure nodes later when the network size expands.

Planning Guidelines for Hierarchical Networks

When you have a plan for dividing your network into multiple peer groups, the next step is to plan communications between those peer groups. To enable communications between peer groups, you will need to identify a peer group leader for each peer group. The following sections provide planning guidelines for the peer group leaders and border nodes in a hierarchical network.

Planning Guidelines for Peer Group Leaders

Peer group leaders are the logical nodes that represent their peer group at the next higher level in the PNNI hierarchy. Peer group leaders are introduced in "Introduction to PNNI." The following are some planning guidelines for peer group leaders:

  • For a peer group to communicate with another peer group, the peer group must have at least one node that is capable of acting as peer group leader.
  • It is good design practice to configure multiple nodes within a peer group to serve as peer group leader. The peer group priority is a configurable parameter that determines which of the peer group leader candidates becomes peer group leader.
  • To compensate for the additional processing requirements of peer group leaders, consider reducing the traffic load for the peer group leader switch and avoid using the same switch as both a peer group leader and a border node.

Planning Guidelines for Border Nodes

Border nodes are physical nodes that are members of one peer group and have connections to member nodes of other peer groups. Border nodes are introduced in "Introduction to PNNI." To compensate for the additional processing requirements of border nodes, consider reducing the intra-peer-group traffic load for the border node and avoid using the same switch as both a border node and a peer group leader.