Downloads |
Table Of Contents
MPF for Broadband LAC, LNS, and PTA
Prerequisites for MPF for Broadband LAC, LNS, and PTA
Restrictions for MPF for Broadband LAC, LNS, and PTA
Information About MPF for Broadband LAC, LNS, and PTA
MPF for Broadband LAC, LNS, and PTA Overview
MPF Accelerated, Punted, and Ignored Features
Port Adapter Coexistence with MPF
MPF-Supported Control Plane Features
Unsupported Cisco IOS Counters
Configuring MPF for Broadband LAC, LNS, and PTA
Disabling MPF for Broadband LAC, LNS, and PTA
Enabling MPF for Broadband LAC, LNS, and PTA
Verifying MPF for Broadband LAC, LNS, and PTA Is Enabled
Configuration Examples for MPF for Broadband LAC, LNS, and PTA
Example of show version Command
Example of show mpf cpu Command
Example of show mpf cpu history Command
Example of show mpf punt Command
Example of show ip interface Command
Example of show interface stats Command
MPF for Broadband LAC, LNS, and PTA
The Multi-Processor Forwarding (MPF) for Broadband L2TP Access Concentrator (LAC), L2TP Network Server (LNS), and PPP Termination and Aggregation (PTA) feature significantly improves broadband feature performance by accelerating features with a fast-forwarding method of switching packets on the second CPU.
History for the MPF for Broadband LAC, LNS, and PTA Feature
Finding Support Information for Platforms and Cisco IOS Software Images
Use Cisco Feature Navigator to find information about platform support and Cisco IOS software image support. Access Cisco Feature Navigator at http://www.cisco.com/go/fn. You must have an account on Cisco.com. If you do not have an account or have forgotten your username or password, click Cancel at the login dialog box and follow the instructions that appear.
Contents
•
Prerequisites for MPF for Broadband LAC, LNS, and PTA
•
•
•
•
Prerequisites for MPF for Broadband LAC, LNS, and PTA
•
–
–
•
–
–
•
•
–
–
–
Restrictions for MPF for Broadband LAC, LNS, and PTA
•
•
•
Enhanced ATM port adapters:
–
–
–
–
–
–
Fast Ethernet/Gigabit Ethernet LAN port adapters:
–
–
–
See the "Port Adapter Coexistence with MPF" section for more information.
•
•
CBAC is only supported for Virtual Access Interfaces (VAIs) and cannot be configured for any MPF-accelerated interfaces. MPF does not accelerate any CBAC features. Instead MPF punts any CBAC-configured VAI session traffic to Cisco IOS for CBAC feature processing. Due to this design limitation, CBAC security ACL filters can only be applied to VAIs and non-MPF port adapter interfaces. CBAC ACL filters cannot be applied to any MPF-accelerated, native GE interfaces.
In addition, the CBAC "pin-hole" feature that allows specified inspect traffic to flow through the CBAC security ACL filters does not work on MPF-accelerated interfaces.
Information About MPF for Broadband LAC, LNS, and PTA
To understand MPF for Broadband LAC, LNS, and PTA, you should read the following sections:
•
•
MPF for Broadband LAC, LNS, and PTA Overview
MPF for Broadband LAC, LNS, and PTA is a method of accelerating a subset of broadband aggregation features on the Cisco 7301 and Cisco 7200 VXR routers by enabling fast-forwarding software on the second CPU. MPF for Broadband LAC, LNS, and PTA significantly improves performance by at least two times that of a regular Cisco 7301 or Cisco 7200 VXR router without any hardware changes.
MPF for Broadband LAC, LNS, and PTA is accomplished by the second CPU, CPU1, running fast-forwarding (FF) software to switch packets. Standard Cisco IOS features use a single CPU, CPU0, for both control plane and data functionality. MPF has moved the data plane traffic to the second CPU where LAC features, and LNS and PTA applications are processed at an accelerated rate, thus improving the performance of these broadband features. The MPF microcode running in CPU1 forwards traffic at approximately twice the forwarding performance of standard Cisco IOS software, allowing up to 1-Gbps line rate throughput. MPF can run all basic LAC, LNS, and PTA functionality with minimal impact to configuration and to Cisco IOS Release 12.3T and later functionality.
All control plane and data plane traffic is fast-forwarded to CPU1 for processing. A subset of MPF features is accelerated. Some non-MPF features that are not accelerated are punted to CPU0 where the non-MPF features are processed by standard Cisco IOS software running on CPU0. Some non-MPF features are considered unsupported and ignored in an MPF system.
When the Cisco IOS image containing MPF is loaded, fast-forwarding is enabled by default and LAC, LNS, and PTA features are accelerated. The first CPU, CPU0, always comes up first, and then the second CPU, CPU1, comes up. MPF is best deployed in networks requiring only MPF accelerated and punted features. The MPF image should not be used as a general purpose router image.
The Cisco IOS Release 12.4(4)T image is the recommended image for deploying MPF.
The NPE-G1 is required for the Cisco 7200 VXR router. LAC, LNS, and PTA functionality for both the Cisco 7301 and Cisco 7200 VXR routers is supported only on native Gigabit Ethernet (GE) ports. Traffic forwarding using the second CPU is only supported on the native GE ports. For example, the three native GE interfaces (G0/0, G0/1 and G0/2) on the Cisco 7301 router use the second CPU for traffic forwarding.
No port adapter traffic is accelerated by the second CPU. Any traffic from these port adapters will be forwarded using just the first CPU. We recommend you migrate any existing port adapter traffic to the native GE ports to take full advantage of the MPF accelerated features.
The benefits of improved network performance are important due to the rapid increase in broadband users. Existing customers who deploy an LNS network can gain performance without adding additional hardware into their network. Service providers can offer services to more subscribers while maintaining their current network topology. Customers who want to add LNS functionality can purchase the LNS MPF image on a new Cisco 7301 or Cisco 7200 VXR NPE-G1 router.
Figure 1 Typical LAC/LNS Network Topology
Figure 1 depicts a typical LAC/LNS solution deployment. In Figure 1, both wholesale and retail broadband models are supported. Either GE or ATM are on the LAC access side and connect to either a GE-DSL Access Multiplexer (DSLAM) or ATM-DSLAM, respectively. MPF accelerated features are only supported for the Cisco 7301 or Cisco 7200 VXR NPE-G1 native GE ports, which require the service provider and Internet core to be Ethernet-based. ATM and MPLS are not accelerated on the LAC side.
The fast-forwarding software is bundled together with the Cisco IOS software image. When the Cisco IOS image is loaded, the second CPU is enabled by default. To disable fast-forwarding, use the no ip mpf command. In addition, show and debug commands monitor forwarding on the second CPU and provide statistics.
The MPF for Broadband LAC, LNS, and PTA feature requires the purchase of software to enable the second CPU. You may purchase the special image when you purchase a new Cisco 7301 or Cisco 7200 VXR router, or you may purchase the special MPF image as an upgrade. Contact your Cisco field representative or sales support team for more information.
MPF Accelerated, Punted, and Ignored Features
When a system is enabled for MPF by enabling fast forwarding on the second CPU (CPU1), all features are fast forwarded to CPU1. All MPF-supported features, comprising a subset of Cisco IOS features, are accelerated. See the "MPF Accelerated Features" section for a list of accelerated features.
Another subset of Cisco IOS features that is not accelerated by MPF and considered non-MPF features is punted to the first CPU for standard Cisco IOS processing. See the "Punted Features" section for a list of punted features.
Some non-MPF features may not undergo correct packet switching and are considered ignored. See the "Ignored MPF Features" section for a list of ignored features.
Port Adapter Coexistence with MPF
MPF does not accelerate any port adapter traffic. Because the first CPU, CPU0, controls all port adapters, there is no MPF accelerated feature support for any traffic that is switched to and from port adapters.
Port adapters generally can functionally coexist with the MPF architecture and are supported by MPF from the standpoint of functionality and stability. Port adapters operate normally within standard Cisco IOS software and port adapter traffic is directly controlled by Cisco IOS.
Certain port adapters are considered MPF-supported. For a list of the supported port adapters, see the "Restrictions for MPF for Broadband LAC, LNS, and PTA" section.
If you switch traffic from a native MPF-accelerated GE port to a port adapter, or from a port adapter to a native MPF-accelerated GE port, standard Cisco IOS performance is not guaranteed. We recommend you migrate any existing port adapter traffic to the native Gigabit Ethernet interfaces to gain the full benefit of MPF accelerated features.
If all port adapter traffic is isolated to CPU0, the native GE port-to-GE port fast-forwarding performance is not affected.
Note
MPF Accelerated Features
MPF accelerated features are a subset of Cisco IOS features. They are Cisco IOS data plane features that are fast forwarded to the second CPU, CPU1, where packet switching is accelerated by MPF fast forwarding software.
Where applicable, all MPF accelerated features can be applied to interfaces, subinterfaces, Remote Authentication Dial-In User Service (RADIUS) per-user profiles, and per-user virtual templates (VTs), unless noted otherwise. RADIUS per-user profile features generally implement global policy maps or use access-group VSAs, that are defined locally on the router, to allow session scalability.
Although a RADIUS per-user profile can specify a customized, per-user Access Control List (ACL) that does not use any locally defined access group on the router, this approach is not recommended due to scaling and performance issues.
Additionally, all existing Cisco IOS Release 12.3(14)T or later CLI syntax is preserved for all features, whether they are MPF accelerated or not.
Note
Table 1 describes the MPF accelerated features. Because these features are already documented in existing Cisco IOS releases or in technical notes, a link to the documentation is provided where relevant.
Table 1 MPF Accelerated Features
Supported encapsulations are:
•
•
•
•
•
•
•
Support for L2TP, PPPoEoE, PPPoEoVLAN, 802.1Q (IP), and 802.3 ARPA ETH encapsulations was introduced in Cisco IOS Release 12.3(7)XI1.
Support for LLC/SAP and LLC/SNAP encapsulations was introduced in Cisco IOS Release 12.3(14)YM2.
L2TP access concentrator (LAC) support
Support was introduced in Cisco IOS Release 12.3(7)XI1.
LAC is a device that is typically (although not always) located at a service provider's point-of-presence (POP), where initial configuration and ongoing management are done by the service provider. Packets sent through tunnels to and from L2TP are accelerated.
IPv4 forwarding
Support was introduced in Cisco IOS Release 12.3(7)XI1.
IP fragmentation
Support was introduced in Cisco IOS Release 12.3(7)XI1.
For more information, see the IP Fragmentation and PMTUD white paper.
L2TP network server (LNS)
Packets sent to the LNS are accelerated. LNS is the termination point for an L2TP tunnel. The LNS initiates outgoing calls and receives incoming calls from the LAC.
For more information, see the following:
•
Managed LNS/Virtual Home Gateway (VHG)
VPN Routing and Forwarding (VRF) Lite support only (no MPLS tag switching).
VRF Lite is a feature that enables a service provider to support two or more VPNs (using only VRF-based IPv4), where IP addresses can be overlapped among the VPNs.
Managed VRF is a new VRF functionally in broadband that allows a wholesale access provider to offer Virtual Private Networks to each respective Internet Service Provider (ISP) customer or corporate customer using not only the access provider's shared IP or MPLS infrastructure, but also using the same Cisco 7200 VXR or Cisco 7401ASR providing LNS or PTA functionality.
PPP Termination and Aggregation (PTA)
PTA is a PPP selection method in which service selection is based on a structured domain name (for example, username@service.com). PTA terminates the PPP session into a single routing domain. Users can only access one service and do not have access to the default network.
For applicable information, see the "Service Selection Methods" chapter of the Cisco 10000 Series Router Service Selection Gateway Configuration Guide.
Managed PTA
VPN Routing and Forwarding (VRF) Lite support only (no MPLS tag switching).
Access Control Lists (ACLs)
ACLs can only be applied at the following:
•
•
•
The show ip access-list command is the only supported CLI.
See the "Access Control Lists Support" section for types of ACLs accelerated, restrictions, and detailed support information.
The following QoS features and Modular QoS CLI (MQC) have limited support:
•
For information on the Single-Rate Policer, see the Traffic Policing feature module.
•
Only QoS classification, policing, and packet marking using the policy map name at the global level is supported.
Policing on an individual interface is supported for the following:
•
•
•
The following are not supported:
•
•
•
•
Limited support. Only the following QoS packet marking policy map actions are supported:
•
•
•
Note
All other set CLI commands are not supported.
Limited support. Only the following QoS classification CLI commands to configure a class map for a per-session policy map are supported:
•
•
•
•
•
All other match commands are not supported.
Refer to the Cisco IOS Quality of Service Solutions Command Reference for information on the match commands
Tunnel Switching
Tunnel switching allows the LAC or LNS tunnel switching device to terminate tunnels from LACs and forward the sessions through new L2TP tunnels selected independently of the client-supplied domains.
For applicable information, see the L2TP Tunnel Switching feature module, Cisco IOS Release 12.1(1)DC.
Multipath Load Balancing
This feature was introduced in Cisco IOS Release 12.3(14)YM2.
Enabled for:
•
•
See the "Multipath Load Balancing" section for more information.
Unicast Reverse Path Forwarding (uRPF)
•
•
•
–
–
–
uRPF allows the router to see if any IP packet received at a router interface arrives on the best reverse path (return route) to the source address of the packet.
See the Unicast Reverse Path Forwarding feature module, Cisco IOS Release 11.1.
Also, see the "Configuring Unicast Reverse Path Forwarding" chapter of the Cisco IOS Security Configuration Guide, Release 12.2.
IP Reassembly
MPF supports IPv4 reassembly of L2TP tunnels and all IPv4 traffic using the second CPU to perform fast-forwarding reassembly of L2TP-encapsulated packets. The MPF reassembly path handles all fragmented IPv4 traffic that is destined for the router, whether it is L2TP traffic or not.
For more information, see the IP Fragmentation and PMTUD white paper.
Jumbo frames
Support for jumbo frames that have a maximum transmission unit (MTU) size greater than 1500 bytes on Ethernet and Gigabit Ethernet interfaces. Larger MTUs can enhance performance by eliminating fragmentation.
For applicable information, see the "Configuring Jumbo Frame Support" section of the Cisco 7600 Series Cisco IOS Software Configuration Guide, 12.2SX.
ip tcp adjust-mss command function
Command behaves as a standard Cisco IOS command.
The ip tcp adjust-mss command must be applied at:
•
•
•
For more information, see the ip tcp adjust-mss command in the "Broadband Access: PPP and Routed Bridge Encapsulation Commands" chapter of the Cisco IOS Wide-Area Networking Command Reference, Release 12.2T.
l2tp ip tos reflect command function
Use in VPDN group configuration mode. Command behaves as a standard Cisco IOS command does.
For more information, see the l2tp ip tos reflect command in the Cisco IOS Dial Technologies Command Reference.
The closely-related commands, ip tos and ip precedence, are documented in the Cisco IOS Bridging and IBM Networking Command Reference.
PPP Keep-alives
•
•
IPv4 ICMP
•
Access Control Lists Support
When MPF is enabled, Turbo Access Control Lists is enabled by default. If MPF is disabled, all ACL packets are punted for Cisco IOS processing on all interfaces.
Table 2 describes the support for different types of access control lists (ACLs) in an MPF network and whether the ACL is accelerated or not supported by MPF and thus ignored. The table lists ACLs by the type of ACL, such as time-based ACL, or the different methods of configuring ACLs, such as with QoS classification commands or configured with or without a RADIUS profile.
Table 2 ACL Support in an MPF Network
Security ACL with IP L3/L4 ACEs
Yes
No
Security ACL with any other format
No
Yes
QoS classification CLI commands
Yes, with restrictions.
For details on restrictions and QoS classification CLI supported, see the Notes and Restrictions column for the "The following QoS features and Modular QoS CLI (MQC) have limited support:" row in Table 1.
No, if restrictions are complied with.
Any other QoS classification configuration CLI
No
Yes
Security ACL logging
Yes
No
Time-based ACL
No
Yes
Reflexive ACL
No
Yes
Virtual template-based per-session or per-user security ACL, configured with QoS classification
Yes, with restrictions.
For details on restrictions and QoS classification CLI supported, see the Notes and Restrictions column for the "The following QoS features and Modular QoS CLI (MQC) have limited support:" row in Table 1.
No, if restrictions are complied with.
Context-based Access Control (CBAC) user ACLs (RADIUS profile configured)
No
Yes
These packets are punted to Cisco IOS for processing.
CBAC user ACLs (non-RADIUS profile configured)
No
Yes
These packets are punted to Cisco IOS for processing.
Other RADIUS ACLs
No
Yes
RADIUS profile configured QoS using ACL (called attribute 11 access-group support)
Yes, with restrictions.
For details on restrictions and QoS classification CLI supported, see the Notes and Restrictions column for the "The following QoS features and Modular QoS CLI (MQC) have limited support:" row in Table 1.
No, if restrictions are complied with.
Marking, after QoS classification
Yes, with restrictions.
For details on restrictions and QoS classification CLI supported, see the Notes and Restrictions column for the "The following QoS features and Modular QoS CLI (MQC) have limited support:" row in Table 1.
No, if restrictions are complied with.
ACLs with unsupported Access Control Entries (ACEs) (such as L2 MAC-address, MPLS EXP)
No
Yes
ACLs on port adapters
No
Yes
Cisco IOS directly handles all port adapter features. All MPF ACL and QoS restrictions do not apply to port adapters.
Multipath Load Balancing
The MPF Multipath Load Balancing feature optimizes resource usage by distributing traffic over multiple paths to transfer data to a destination when more than one path to the same destination is available.
Multipath Load Balancing supports per-destination load balancing, but not per-packet load balancing. Load balancing decisions are made on the outbound interface. When you configure load balancing, configure it on outbound interfaces. Per-destination load balancing is enabled by default when MPF is enabled. Disabling MPF disables MPF Multipath Load Balancing.
Multipath Load Balancing decides which path (or adjacency) to use, based on the source and destination IP addresses and the IP field of the packet. Network nodes in the network are said to be adjacent if they can reach each other with a single hop across a link layer. If only IP information is considered, packets with the same source and destination IP address pair are forwarded out of the same path. However, packets with different source and destination IP addresses typically go out different paths.
Per-Destination Load Balancing
Per-destination load balancing allows the router to use multiple paths to achieve load sharing. Packets for a given source-destination host pair are guaranteed to take the same path, even if multiple paths are available. Traffic destined for different pairs tends to take different paths.
Per-destination load balancing ensures that packets destined for a given host pair arrive in order and are routed over the same links. Load sharing becomes more effective as the number of source-destination pairs increase because per-destination load balancing depends on the statistical distribution of traffic.
Multipath Load Balancing Algorithm
The data used to make a Multipath Load Balancing decision is passed through a hash function. The router uses the result of the hash function to pick the path or adjacency to load share. The hash function is based on the Bob Jenkins' algorithm. This algorithm is different from the one used by CEF Load Balancing and is found at the following URL: http://burtleburtle.net/bob/hash/evahash.html.
Restrictions
Multipath Load Balancing does not support port adapter interfaces.
Traffic is not load-balanced when traffic is sent between a native Gigabit Ethernet interface and a port adapter interface. Only 25% of the traffic is sent through the port adapter.
The traffic is equally divided at input, so that 50% of the traffic is MPF accelerated and is load-balanced via Multipath Load Balancing. The other 50% of the traffic is punted. The 50% that is punted is processed by standard Cisco IOS CEF load balancing, resulting in 25% of the traffic going to the port adapter and 25% going to the Gigabit Ethernet interface.
Multipath Load Balancing Command
The following command is supported to show which routing decision (next hop) is made for a given IP address pair:
show mpf ip exact-route [vrf vrf_name] src-ip-addr dst-ip-addr
See the show mpf ip exact-route command reference page in this document for more information.
MPF-Supported Control Plane Features
The first CPU continues to manage the control plane features, which are typically not accelerated. MPF supports the standard control plane features in Cisco IOS Release 12.3(14)T and later. Table 3 describes the MPF-supported control plane features and any exceptions. The exception is that although Cisco IOS data counters are supported, protocol-specific counters and class map counters are not supported.
MPF accelerated features are compatible with the control plane features in providing accurate information or proper data plane configuration setup.
Table 3 MPF-Supported Control Plane Features
Logical line ID (LLID) blocking
See "Related Documents" section for documentation information.
Per VRF AAA
Domain to VRF Mapping
MTU Adjust features
The following are not MPF-supported:
•
•
L2TP Load Balancing with Random Tunnel Selection
Support was introduced in Cisco IOS Release 12.4(4)T. For more information, see the "Configuring L2TP Tunnel Server Load Balancing and Failover Using the RADIUS Tunnel Preference Attribute" section in "How to Configure AAA for VPDNs" in the Cisco IOS VPDN Configuration Guide.
DHCP Option 82
Supported for:
•
•
•
•
SNMP
MIBs
CISCO-PROCESS-MIB was modified to retrieve statistics for both the first CPU (CPU0) and second CPU (CPU1). CPU0 always comes up first, and then CPU1. In the cpmCPUTotalTable, the second entry is statically reserved for MPF CPU1; the first entry is IOS CPU0. See the "MIBs" section for more information.
Cisco IOS Statistics—all Cisco IOS counters are supported, with exceptions.
The exceptions are that protocol-specific counters, class map counters, and class-default counters are not supported. See the "Unsupported Cisco IOS Counters" section for more information.
Cisco IOS show commands
All standard RADIUS Accounting is supported.
All standard RADIUS Attributes are supported.
SYSLOG messages
ACL logs
Unsupported Cisco IOS Counters
In an MPF system, when a show policy-map interface command is issued, the class map counters display zero packets matched, even though packets are classified.
In the following example, policing and packet marking are configured in the same class map. Issuing the show policy-map interface command displays "0" matches, but shows packets being marked. The bold-faced text is for documentation purposes only.
Router# show policy-map interfaceGigabitEthernet0/1Service-policy input: pmap2Class-map: cmap3 (match-any)0 packets, 0 bytes5 minute offered rate 0 bps, drop rate 0 bps =>MPF does not update class map countsMatch: access-group 210 packets, 0 bytes5 minute rate 0 bps ======>MPF does not update match countspolice:cir 8000 bps, bc 1500 bytesconformed 15 packets, 1500 bytes; actions:transmitexceeded 0 packets, 0 bytes; actions:dropconformed 0 bps, exceed 0 bps ======>MPF and Cisco IOS update policer countsQoS Setprecedence 5Packets marked 15 ======>MPF and Cisco IOS update marking countsClass-map: class-default (match-any)0 packets, 0 bytes5 minute offered rate 0 bps, drop rate 0 bpsMatch: any ======>MPF does not update class-default countsPunted Features
Packets of features that are not accelerated MPF features are punted to CPU0 where the packets are processed by standard Cisco IOS software. Punted packets can be either control plane or data plane features.
When a punted packet type is received by MPF on CPU1, MPF sends the packet to CPU0 using a priority queue, where packets are flow-controlled via a credit-based design. Control plane packets are given higher priority over data plane packets.
Punted packets are completely handled by Cisco IOS software for all ingress and egress features and are viewed by Cisco IOS as packets received by the Cisco IOS port adapter driver.
Note
In the case of a port adapter-to-port adapter switching path or a hybrid switching path where one interface is a native MPF accelerated GE interface and the other is a port adapter running Cisco IOS (for example, ATM), the packets are punted to CPU0 for standard Cisco IOS feature processing of both ingress and egress features. Therefore your system gains no MPF performance improvements when you deploy port adapter-to-port adapter traffic switching as opposed to deploying MPF accelerated, native GE interface-to-GE interface traffic switching. See the "Port Adapter Coexistence with MPF" section for more information on port adapters.
Table 4 describes the punted features in an MPF system.
Table 4 Punted Features
Multicast for IPv4 and IPv6
Multicast User Authentication and Profile Support (IPv4 and IPv6)
Multicast User Authentication and Profile Support for IPv6 was introduced in Cisco IOS Release 12.4(4)T. For more information, see the Implementing IPv6 Multicast module in the Cisco IOS IPv6 Configuration Guide.
IPv6 Unicast
VAI Context-based Access Control (CBAC)
L2TP Keep-Alives (Hello request/replies)
Packets destined to router, except for MPF accelerated features (LNS packets)
Broadcast packets (including DHCP)
CEF Accounting
ARP
ARP packets are punted at high priority.
PPP IP Control/Unaccelerated PPP Datagrams including but not limited to:
•
•
•
•
•
•
Excluding IP datagram and PPP echo-replies
Ignored MPF Features
Ignored features in an MPF network are those Cisco IOS features that are neither accelerated by MPF nor punted for Cisco IOS processing. When an ignored MPF feature is applied to a native GE interface on an MPF system, the system accepts the configuration. However the unsupported MPF feature is ignored for traffic going through that native GE interface.
An ignored MPF feature is not punted and not processed by Cisco IOS software. The feature is ignored and does not cause any unexpected Cisco IOS drops, crashes, network instability, or major behavioral differences.
Ignored Warning Message
When an ignored MPF feature is configured, the system issues a console warning message for main physical interfaces and virtual templates. To avoid console loading issues, a single warning message is issued only for the first occurrence of the specific ignored feature for subinterfaces (VLANs) and virtual accesses (sessions).
Depending on whether the ignored feature is configured on the input or output direction for the interface or both, the warning message notes the configured input or output feature.
An example of the warning message follows:
Router#%MPF-4-IGNOREDFEATURES:Interface Gi0/3:Input "PBR" configurations arenot MPF supported and are IGNORED.%SYS-5-CONFIG_I:Configured from console by consoleRouter#Router(config-if)#%MPF-4-IGNOREDFEATURES:Interface Gi0/3:Input "PBR" configurations arenot MPF supported and are IGNORED.%MPF-4-IGNOREDFEATURES:Interface Gi0/3:Output "NetFlow" configurationsare not MPF supported and are IGNORED.Ignored Warning for Class-Map Feature Configuration—Tip
The following tip advises you of MPF behavior. When you make a policy configuration change to an already configured class-map feature definition, MPF issues an ignored warning message for any QoS feature that is ignored by MPF, even though you are not changing the QoS feature.
The following example shows an existing policy as configured on Gigabit Ethernet0/2 and the resulting ignored warning message when a change in precedence value is made to class-map 2:
Existing policy on Gigabit Ethernet0/2 is:
class-map match-all Class1match anyclass-map match-all Class2match anypolicy-map Policy1 class Class1set precedence 2 set qos-group 5 class Class2set precedence 0 set qos-group 5You change the precedence value for class-map 2 from a value of 0 to 2, which MPF processes. However, the resulting ignored warning message is issued, even though you are not actually changing the QoS group feature:
Router(config)# policy-map Policy1Router(config-pmap)# class Class2Router(config-pmap-c)# set precedence 2*Aug 3 01:58:37.184: %MPF-4-NOFEATURESUPPORT: Interface Gi0/2: QOS SET:'set local QoS-group' is an unsupported MPF feature and is IGNORED.Major Ignored MPF Features
Many of the ignored MPF features are related to QoS policing and QoS packet marking. See Table 5 for a list of some of the major ignored MPF features. Note that this is not a comprehensive list of all Cisco IOS features that are not supported by MPF.
Table 5 Major Ignored MPF Features
The following encapsulations are not supported:
•
•
Layer 2 Forwarding (L2F) protocol
The following QoS features and Modular QoS CLI (MQC) are not supported:
•
•
•
•
•
•
•
•
•
•
•
•
uRPF Suppressed Drop ACLs
ACLs:
•
•
•
•
See the "Access Control Lists Support" section for additional information on unsupported ACLs.
Network Address Translation (NAT)
Network-Based Application Recognition (NBAR)
Virtual Template Pre-Cloning
The Cisco IOS Release 12.3(14)T CLI is not officially supported. We strongly recommend that you do not enable this feature because it disables the Cisco IOS idb scaling and session performance enhancement features.
System Memory Requirements
The second CPU, CPU1, uses shared system memory for various accelerated features. Table 6 describes the minimum system memory requirements for all supported routers by type of deployment and whether the deployment is "basic" or features are turned on.
Basic Deployment
In a basic deployment, there are no configured features and only minimal routes. Specifically there is no use of ACLs, QoS Policers, or VRF instances. The maximum number of required routes fit into the available memory space.
Features Deployment
In a features deployment, any features and number of routes can be configured.
MPF Scalability
MPF supports the same scaling limits set by the Cisco 7200 series and Cisco 7301 routers.
MPF accelerated features are subject to the same scaling limits imposed by Cisco IOS Release 12.3T and do not attempt to increase the actual customer-usable scalability limits.
Configuring MPF for Broadband LAC, LNS, and PTA
This section contains the following tasks:
•
•
•
Disabling MPF for Broadband LAC, LNS, and PTA
This section contains the procedure to disable fast-forwarding on the second CPU on a Cisco 7301 or Cisco 7200 VXR router for MPF for Broadband LAC, LNS, and PTA.
Prerequisites
You have installed the MPF software image.
SUMMARY STEPS
1.
2.
3.
4.
DETAILED STEPS
Enabling MPF for Broadband LAC, LNS, and PTA
This section contains the procedure to re-enable fast-forwarding on the second CPU on a Cisco 7301 or Cisco 7200 VXR router for MPF for Broadband LAC, LNS, and PTA.
Prerequisites
•
•
SUMMARY STEPS
1.
2.
3.
4.
5.
DETAILED STEPS
Verifying MPF for Broadband LAC, LNS, and PTA Is Enabled
Perform this optional task to verify that MPF fast-forwarding is enabled for the second CPU.
SUMMARY STEPS
1.
2.
DETAILED STEPS
Configuration Examples for MPF for Broadband LAC, LNS, and PTA
This section contains examples of the following show commands used to verify MPF configurations.
•
•
•
•
•
•
Example of show version Command
The following show version command example displays the second CPU (CPU1) enabled for MPF for Broadband LAC, LNS, and PTA. Note that the show version output displays "SB-1 CPU" when processor 1 is enabled. Some of the output is set in boldface for documentation purposes to show that the MPF software is present.
Cisco 7301 show version Command Example
Router# show versionCisco IOS Software, 7301 Software (C7301-I12S-M), Experimental Version 12.3(20040524:050554) [REL-v123_7_xi_throttle.ios-weekly 114]Copyright (c) 1986-2004 by Cisco Systems, Inc.Compiled Mon 24-May-04 06:26 by useridROM:System Bootstrap, Version 12.3(4r)T2, RELEASE SOFTWARE (fc1)BOOTLDR:Cisco IOS Software, 7301 Software (C7301-BOOT-M), Experimental Version 12.3(20040514:051116) [userid-v123_7_xi_throttle-ios-nightly-task 117]Router uptime is 1 hour, 50 minutesSystem returned to ROM by reload at 18:05:37 UTC Wed Mar 22 2004System image file is "disk0:c7301-i12s-mz"Cisco 7301 (NPE) processor (revision C) with 229376K/32768K bytes of memory.Processor board ID 74806813SB-1 CPU at 700MHz, Implementation 1, Rev 0.2, 512KB L2 Cache1 slot midplane, Version 3.0Last reset from watchdog nmiCPU 1 Multi-Processor Forwarding, Fri May 21 14:21:57 2004 [dailybuil 119]1 FastEthernet interface3 GigabitEthernet interfaces509K bytes of NVRAM.62976K bytes of ATA PCMCIA card at slot 0 (Sector size 512 bytes).32768K bytes of Flash internal SIMM (Sector size 256K).Configuration register is 0x0Cisco 7200 VXR show version Command Example
Router# show versionCisco IOS Software, 7200 Software (C7200-I12S-M), Version 12.3(BSNOP4_NIGHTLY.050202) UBUILDIT Image, CISCO DEVELOPMENT TEST VERSIONCopyright (c) 1986-2005 by Cisco Systems, Inc.Compiled Wed 02-Feb-05 04:57 byROM: System Bootstrap, Version 12.3(4r)T3, RELEASE SOFTWARE (fc1)BOOTLDR: Cisco IOS Software, 7200 Software (C7200-KBOOT-M),12.3(20040712:050512) [REL-v123_7_xi_throttle.ios-weekly 115]Lac1 uptime is 16 hours, 22 minutesSystem returned to ROM by reload at 18:12:49 UTC Mon Feb 7 2005System image file is "tftp://223.255.254.254/auto/tftpboot-users/agiyer/c7200-i12s-mz.2005-02-02.BSNOP4_NIGH"Last reload reason: Reload commandCisco 7206VXR (NPE-G1) processor (revision 0xFF) with 983040K/65536K bytes of memory.Processor board ID 31650243SB-1 CPU at 700MHz, Implementation 1025, Rev 0.2, 512KB L2 Cache6 slot VXR midplane, Version 2.7Last reset from power-onCPU 1 Multi-Processor Forwarding, Tue Feb 1 00:48:42 PST 2005 [userid 100]PCI bus mb1 (Slots 1, 3 and 5) has a capacity of 600 bandwidth points.Current configuration on bus mb1 has a total of 400 bandwidth points.This configuration is within the PCI bus capacity and is supported.PCI bus mb2 (Slots 2, 4 and 6) has a capacity of 600 bandwidth points.Current configuration on bus mb2 has a total of 700 bandwidth points.The set of PA-2FE, PA-POS-2OC3, and I/O-2FE qualify for "halfbandwidth points" consideration, when full bandwidth point countingresults in oversubscription, under the condition that only one of thetwo ports is used. With this adjustment, current configuration on busmb2 has a total of 700 bandwidth points.This configuration has oversubscripted the PCI bus and is not asupported configuration.Please refer to the following document "Cisco 7200 Series PortAdaptor Hardware Configuration Guidelines" on CCO <www.cisco.com>,for c7200 bandwidth points oversubscription/usage guidelines.WARNING: PCI bus mb2 Exceeds 600 bandwidth points1 Ethernet interface2 FastEthernet interfaces5 Gigabit Ethernet interfaces1 ATM interface509K bytes of NVRAM.62976K bytes of ATA PCMCIA card at slot 2 (Sector size 512 bytes).16384K bytes of Flash internal SIMM (Sector size 256K).Configuration register is 0x2Example of show mpf cpu Command
Use the show mpf cpu command to display the average second CPU utilization in the last five seconds, one minute, and five minutes.
router# show mpf cpuCPU utilization for five seconds: 33%; one minute: 25%; five minutes: 30%Example of show mpf cpu history Command
Use the show mpf cpu history command to graph output of the second CPU utilization for the last 60 seconds, 60 minutes, and 72 hours.
Router# show mpf cpu historyslns 12:12:40 AM Saturday Nov 18 2000 UTC 3333333333333333333333333333333333333333333333333333333333 333333333333333333333333333333333333333333333333333333333310090807060504030 ***************************20 ***************************10 ***************************0....5....1....1....2....2....3....3....4....4....5....5....0 5 0 5 0 5 0 5 0 5CPU% per second (last 60 seconds) 3333333333333333333333333333333333333333333333333333333333 333333333333333333333333333333333333333333333333333333333310090807060504030 #################20 #################10 #################0....5....1....1....2....2....3....3....4....4....5....5....0 5 0 5 0 5 0 5 0 5CPU% per minute (last 60 minutes)* = maximum CPU% # = average CPU%16080100 *90 *80 *70 **60 **50 **40 ##30 ##20 ##10 ##0....5....1....1....2....2....3....3....4....4....5....5....6....6....7.0 5 0 5 0 5 0 5 0 5 0 5 0CPU% per hour (last 72 hours)* = maximum CPU% # = average CPU%Example of show mpf punt Command
Use the show mpf punt command to display the per-box punt reason and punt packet counts.
Router# show mpf puntslns#show mpf puntType Message Countl2tp unknown session errors 7l2tp L2TP control 6ipv4/verify adjacency punt 1ethernet unknown ethernet type 542ppp punts due to unknown protocol 333arp ARP request 6Example of show ip interface Command
This example uses the show ip interface command to show interface information on interface Gigabit Ethernet0/3. The example shows that MPF is enabled and the PBR and NetFlow features are not supported by MPF and ignored. The highlighted arrow (for documentation purposes only) shows the configured output and input features and the additional MPF interface information.
Router# show ip interface g0/3GigabitEthernet0/3 is up, line protocol is upInternet address is 155.1.1.1/16Broadcast address is 255.255.255.255Address determined by setup commandMTU is 1500 bytesHelper address is not setDirected broadcast forwarding is disabledOutgoing access list is not setInbound access list is not setProxy ARP is enabledLocal Proxy ARP is disabledSecurity level is defaultSplit horizon is enabledICMP redirects are always sentICMP unreachables are always sentICMP mask replies are never sentIP fast switching is enabledIP fast switching on the same interface is disabledIP Flow switching is disabledIP CEF switching is enabledIP Feature Fast switching turbo vectorIP VPN Flow CEF switching turbo vectorIP multicast fast switching is enabledIP multicast distributed fast switching is disabledIP route-cache flags are Fast, CEFRouter Discovery is disabledIP output packet accounting is disabledIP access violation accounting is disabledTCP/IP header compression is disabledRTP/IP header compression is disabledPolicy routing is enabled, using route map PBRNetwork address translation is disabledBGP Policy Mapping is disabledIP Multi-Processor Forwarding is enabled <======== MPF informationIP Input features, "PBR",are not supported by MPF and are IGNOREDIP Output features, "NetFlow",are not supported by MPF and are IGNOREDExample of show interface stats Command
Use the show interface stats command to display counters for MPF switched packets in an MPF network.
Router# show interface statsGigabitEthernet0/1Switching path Pkts In Chars In Pkts Out Chars OutProcessor 33090 6424353 86353 8645650Route cache 2 116 0 0Multi-Processor Fwding 1004 158632 5004 790632Total 34096 6583101 91357 9436282Additional References
The following sections provide references related to MPF for Broadband LAC, LNS, and PTA.
Related Documents
Cisco 7301 router: Upgrading and Troubleshooting Tasks, using the show version command
Cisco 7301 Installation and Configuration Guide, "Starting and Configuring the Router" chapter
Cisco 7200 VXR routers: NPE-G1 Installation and Configuration Information
Network Processing Engine and Network Services Engine Installation and Configuration, "NPE-G1 Installation and Configuration Information" chapter, "Enabling the Second Processor"
IP commands: command syntax, command mode, defaults, usage guidelines, and examples
Cisco IOS IP Command Reference, Volume 1 of 4: Addressing and Services, Release 12.3T
Logical Line ID (LLID) blocking functionality
See the RADIUS Logical Line ID and L2TP Calling Station ID Suppression feature modules for information on using LLID.
Standards
RFC 2661
L2TP—Internet Engineering Task Force (IETF) standard
Gigabit Ethernet IEEE 802.3z
—
IEEE 802.1Q
—
MIBs
RFCs
No new or modified RFCs are supported by this feature, and support for existing RFCs has not been modified by this feature.
—
Technical Assistance
Command Reference
This section documents only new and modified commands.
New Commands•
•
•
•
•
•
•
Modified Commands•
•
•
clear mpf interface
To clear Multi-Processor Forwarding (MPF) packet counts on all physical interfaces, use the clear mpf interface command in user EXEC or privileged EXEC mode.
clear mpf interface
Syntax Description
This command has no arguments or keywords.
Command Modes
User EXEC
Privileged EXECCommand History
Usage Guidelines
This command has no output. It resets the packet counters shown in the show mpf interface command output.
Examples
The following example uses the clear mpf interface command to reset the packet counters displayed in the output of the show mpf interface command:
Router# clear mpf interfaceRelated Commands
clear mpf punt
To clear Multi-Processor Forwarding (MPF) per-box punt reason and counts, use the clear mpf punt command in user EXEC or privileged EXEC mode.
clear mpf punt
Syntax Description
This command has no arguments or keywords.
Command Modes
User EXEC
Privileged EXECCommand History
Usage Guidelines
This command clears all punt counters and implicitly generates show mpf punt output. It resets for each box or router chassis the punt packet counters shown in the show mpf punt command output. Packets that are punted are directed for Cisco IOS processing and are not accelerated by MPF.
Examples
The following example clears the type of packets (Type), the reasons for the punt (Message), and the punt packet counts (Count) for the router chassis, then implicitly generates show mpf punt output.
Router# show mpf puntType Message Countl2tp unknown session errors 7l2tp L2TP control 6ipv4/verify adjacency punt 1ethernet unknown ethernet type 542ppp punts due to unknown protocol 333arp ARP request 6Router# clear mpf puntType Message Countarp ARP request 38ethernet unknown ethernet type 591l2tp unknown session errors 71790l2tp unsupported output feature 24000Table 7 describes the fields in the clear mpf punt output display.
Related Commands
ip mpf
To enable Multi-Processor Forwarding (MPF) on the second CPU of a Cisco 7200 VXR and Cisco 7301 routers, use the ip mpf command in global configuration mode. To disable MPF, use the no form of this command.
ip mpf
no ip mpf
Syntax Description
This command has no arguments or keywords.
Command Default
MPF is enabled by default on the second CPU.
Command Modes
Global configuration
Command History
Usage Guidelines
This command allows you to disable and reenable MPF.
MPF is enabled by default on the second CPU (CPU1). The special MPF image is bundled together with the Cisco IOS image and must be purchased.
Note
Because MPF is enabled by default when the special MPF image is booted up, if CEF is not enabled, MPF is not enabled and boots up with an error message in the error log.
Examples
The following example disables MPF on the second CPU:
Router(config)# no ip mpfThe following configuration example shows a system where CEF is disabled and the resulting error message showing that MPF cannot be enabled:
00:00:13:%MPF-4-NOIPCEF:MPF disabled due to IP CEF disabled00:00:13:%MPF-6-MODULE:CPU 1 switching module is readyThe following configuration example shows that 1) CEF cannot be disabled until MPF is disabled first; and 2) MPF cannot be enabled until CEF is enabled first:
Router(config)# no ip cef%Cannot disable CEF on this platformRouter(config)# no ip mpfRouter(config)# no ip cefRouter(config)# ip mpf%Can not enable MPF when CEF is disabled.Router(config)# ip cefRouter(config)# ip mpfRelated Commands
show interface stats
To display numbers of packets that were process switched, fast switched, and distributed switched, use the show interface stats command in user EXEC or privileged EXEC mode.
show interface type number stats
Syntax Description
Command Modes
User EXEC
Privileged EXECCommand History
Usage Guidelines
Use this command on the RP.
Note
Examples
The following sample output is from Cisco IOS Release 12.3(14)YM2 and shows counters for both Multi-Processor Forwarding (MPF) switched packets on native GigabitEthernet interfaces and for non-MPF FastEthernet interfaces:
Router# show interface statsGigabitEthernet0/0Switching path Pkts In Chars In Pkts Out Chars OutProcessor 0 0 225 77625Route cache 0 0 0 0Multi-Processor Fwding 950 221250 500 57000Total 950 221250 725 134625GigabitEthernet0/1Switching path Pkts In Chars In Pkts Out Chars OutProcessor 1 60 226 77685Route cache 0 0 0 0Multi-Processor Fwding 500 57000 500 57000Total 501 57060 726 134685GigabitEthernet0/2Switching path Pkts In Chars In Pkts Out Chars OutProcessor 1 60 226 77685Route cache 0 0 0 0Multi-Processor Fwding 0 0 0 0Total 1 60 226 77685FastEthernet1/0Switching path Pkts In Chars In Pkts Out Chars OutProcessor 34015 5331012 1579 158190Route cache 0 0 0 0Total 34015 5331012 1579 158190The following is sample output from the show interface stats command:
Router# show interface fddi 3/0/0 statsFddi3/0/0Switching path Pkts In Chars In Pkts Out Chars OutProcessor 3459994 1770812197 4141096 1982257456Route cache 10372326 3693920448 439872 103743545Distributed cache 19257912 1286172104 86887377 1184358085Total 33090232 2455937453 91468345 3270359086Table 8 describes the significant fields in the display.
show ip interface
To display the usability status of interfaces configured for IP, use the show ip interface command in privileged EXEC mode.
show ip interface [type number] [brief]
Syntax Description
type
(Optional) Interface type.
number
(Optional) Interface number.
brief
(Optional) Displays a summary of the usability status information for each interface.
Command Modes
Privileged EXEC
Command History
Usage Guidelines
The Cisco IOS software automatically enters a directly connected route in the routing table if the interface is usable. A usable interface can send and receive packets. If an interface is not usable, the directly connected routing entry is removed from the routing table. Removing the entry allows the software to use dynamic routing protocols to determine backup routes to the network, if any.
If the interface can provide two-way communication, the line protocol is marked "up." If the interface hardware is usable, the interface is marked "up."
If you specify an optional interface type, you see information for that specific interface.
If you specify no optional arguments, you see information on all the interfaces.
When an asynchronous interface is encapsulated with PPP or Serial Line Internet Protocol (SLIP), IP fast switching is enabled. A show ip interface command on an asynchronous interface encapsulated with PPP or SLIP displays a message indicating that IP fast switching is enabled.
The show ip interface brief command can be used to view a summary of the router interfaces. This command displays the IP address, interface status, and additional information.
Examples
The following examples from Cisco IOS Release 12.3(14)YM2 show:
•
•
The highlighted arrows (for documentation purposes only) show the configured output and input features and the additional MPF interface information.
Router# show running-config interface g 0/3interface GigabitEthernet0/3ip address 10.1.1.1 255.255.0.0ip flow egress <== outputip policy route-map PBR_NAME <== inputduplex autospeed automedia-type gbicnegotiation autoendRouter# show ip interface g 0/3GigabitEthernet0/3 is up, line protocol is upInternet address is 10.1.1.1/16Broadcast address is 255.255.255.255Address determined by setup commandMTU is 1500 bytesHelper address is not setDirected broadcast forwarding is disabledOutgoing access list is not setInbound access list is not setProxy ARP is enabledLocal Proxy ARP is disabledSecurity level is defaultSplit horizon is enabledICMP redirects are always sentICMP unreachables are always sentICMP mask replies are never sentIP fast switching is enabledIP fast switching on the same interface is disabledIP Flow switching is disabledIP CEF switching is enabledIP Feature Fast switching turbo vectorIP VPN Flow CEF switching turbo vectorIP multicast fast switching is enabledIP multicast distributed fast switching is disabledIP route-cache flags are Fast, CEFRouter Discovery is disabledIP output packet accounting is disabledIP access violation accounting is disabledTCP/IP header compression is disabledRTP/IP header compression is disabledPolicy routing is enabled, using route map PBRNetwork address translation is disabledBGP Policy Mapping is disabledIP Multi-Processor Forwarding is enabled <======== MPF informationIP Input features, "PBR",are not supported by MPF and are IGNOREDIP Output features, "NetFlow",are not supported by MPF and are IGNOREDThe following example identifies a downstream VRF. The highlighted line (for documentation purposes only) identifies the downstream VRF.
Router# show ip interface vi 3Virtual-Access3 is up, line protocol is upInterface is unnumbered. Using address of Loopback2 (10.0.0.8)Broadcast address is 255.255.255.255Peer address is 10.8.1.1MTU is 1492 bytesHelper address is not setDirected broadcast forwarding is disabledOutgoing access list is not setInbound access list is not setProxy ARP is enabledLocal Proxy ARP is disabledSecurity level is defaultSplit horizon is enabledICMP redirects are always sentICMP unreachables are always sentICMP mask replies are never sentIP fast switching is enabledIP fast switching on the same interface is enabledIP Flow switching is disabledIP CEF switching is enabledIP Feature Fast switching turbo vectorIP VPN CEF switching turbo vectorVPN Routing/Forwarding "U"Downstream VPN Routing/Forwarding "D"IP multicast fast switching is disabledIP multicast distributed fast switching is disabledIP route-cache flags are Fast, CEFRouter Discovery is disabledIP output packet accounting is disabledIP access violation accounting is disabledTCP/IP header compression is disabledRTP/IP header compression is disabledPolicy routing is disabledNetwork address translation is disabledWCCP Redirect outbound is disabledWCCP Redirect inbound is disabledWCCP Redirect exclude is disabledBGP Policy Mapping is disabledTable 9 describes the significant fields shown in the display.
The following is sample output from the show ip interface brief command:
Router# show ip interface briefInterface IP-Address OK? Method Status ProtocolEthernet0 10.108.00.5 YES NVRAM up upEthernet1 unassigned YES unset administratively down downLoopback0 10.108.200.5 YES NVRAM up upSerial0 10.108.100.5 YES NVRAM up upSerial1 10.108.40.5 YES NVRAM up upSerial2 10.108.100.5 YES manual up upSerial3 unassigned YES unset administratively down down
Related Commands
show mpf ip exact-route
To display the exact route for a source-destination address IP pair in a Multi-Processor Forwarding (MPF) system, use the show mpf ip exact-route command in user EXEC or privileged EXEC mode.
show mpf ip exact-route [vrf vrf-name] src-ip-addr dst-ip-addr
Syntax Description
Command Default
No default behavior or values.
Command Modes
User EXEC
Privileged EXECCommand History
Usage Guidelines
When you are load balancing per destination, this command shows the exact next hop that is used for a given IP source-destination pair.
Examples
The following sample output displays the exact next hop (10.1.104.1) for the specified source IP address (10.1.1.1) and destination IP address (172.17.249.252):
Router# show mpf ip exact-route 10.1.1.1 172.17.249.25210.1.1.1 -> 172.17.249.252 :GigabitEthernet2/0 (next hop 10.1.104.1)Table 11 describes the significant fields shown in the output example.
Related Commands
show mpf cpu
To display the average CPU utilization over a duration of the last 5 seconds, the last 1 minute, and the last 5 minutes when Multi-Processor Forwarding (MPF) is enabled on the second CPU, use the show mpf cpu command in user EXEC or privileged EXEC mode.
show mpf cpu [history]
Syntax Description
history
(Optional) Displays graphical output of the second CPU utilization over the last 60 seconds, the last 60 minutes, and the last 72 hours.
Command Default
No default behavior or values.
Command Modes
User EXEC
Privileged EXECCommand History
Examples
The following example shows that the average utilization of the second CPU is 33 percent for the last 5 seconds, 25 percent for the last minute, and 30 percent for the last 5 minutes:
Router# show mpf cpuCPU utilization for five seconds: 33%; one minute: 25%; five minutes: 30%The following example shows graphical output of utilization of the second CPU for the last 60 seconds (percentage of CPU use per second), the last 60 minutes (percentage of CPU use per minute), and the last 72 hours (percentage of CPU use per hour).
Router# show mpf cpu historyslns 12:12:40 AM Saturday Nov 18 2000 UTC 3333333333333333333333333333333333333333333333333333333333 333333333333333333333333333333333333333333333333333333333310090807060504030 ***************************20 ***************************10 ***************************0....5....1....1....2....2....3....3....4....4....5....5....0 5 0 5 0 5 0 5 0 5CPU% per second (last 60 seconds) 3333333333333333333333333333333333333333333333333333333333 333333333333333333333333333333333333333333333333333333333310090807060504030 #################20 #################10 #################0....5....1....1....2....2....3....3....4....4....5....5....0 5 0 5 0 5 0 5 0 5CPU% per minute (last 60 minutes)* = maximum CPU% # = average CPU%16080100 *90 *80 *70 **60 **50 **40 ##30 ##20 ##10 ##0....5....1....1....2....2....3....3....4....4....5....5....6....6....7.0 5 0 5 0 5 0 5 0 5 0 5 0CPU% per hour (last 72 hours)* = maximum CPU% # = average CPU%Related Commands
show mpf interface
To display Multi-Processor Forwarding (MPF) packet counter information on each physical interface, use the show mpf interface command in user EXEC or privileged EXEC mode.
show mpf interface [interface-name-and-number] [dot1q-vlan-num]
Syntax Description
Command Default
No default behavior or values.
Command Modes
User EXEC
Privileged EXECCommand History
Usage Guidelines
This command is supported for physical interfaces and subinterfaces. There is no support for the virtual access interface (VAI).
You can display the interface count information for a specific Gigabit Ethernet interface by specifying the interface name and number. To display interface information for a specified subinterface only, you must use the 802.1Q VLAN number for the subinterface because the MPF software does not recognize the subinterface number.
Using the show mpf interface command without arguments displays the interface information for all Gigabit Ethernet interfaces and subinterfaces.
Using the clear mpf interface command resets the interface packet counters shown in the show mpf interface command output.
Examples
The following example using the show mpf interface command without arguments displays interface information about up or down state, type of counter (receiving or transmitting packet or bytes), and count number for packets or bytes for all Gigabit Ethernet interfaces (only GigabitEthernet0/1 in this example) and subinterfaces:
Router# show mpf interfaceName Index State Counter CountGi0/1 0 up RX packets 1004RX bytes 158632TX packets 5004Name Index State Counter CountTX bytes 790632RX punts 32961TX punts 85972Gi0/1 1 upGi0/1.100 100 up RX packets 1004RX bytes 158632TX packets 5004TX bytes 790632RX punts 25Gi0/1.101 101 upGi0/1.102 102 upGi0/1.105 105 upGi0/1.106 106 upGi0/1.107 107 upGi0/1.200 200 upGi0/1.201 201 up RX punts 29Gi0/1.202 202 upGi0/1.206 206 upGi0/1.2002 602 up RX punts 26114Gi0/1.2004 604 upThe following example specifies interface information for Gigabit Ethernet interface 0/1 subinterface 100. However, all Gigabit Ethernet interface and subinterface information is displayed because MPF does not recognize the subinterface number, unless it is a VLAN number.
Router# show mpf interface GigabitEthernet0/1.100Name Index State Counter CountGi0/1 0 up RX packets 1004RX bytes 158632TX packets 5004TX bytes 790632RX punts 32996TX punts 86062Gi0/1 1 upGi0/1.100 100 up RX packets 1004RX bytes 158632TX packets 5004TX bytes 790632RX punts 25Gi0/1.101 101 upGi0/1.102 102 upGi0/1.105 105 upGi0/1.106 106 upGi0/1.107 107 upGi0/1.200 200 upGi0/1.201 201 up RX punts 29Gi0/1.202 202 upGi0/1.206 206 upGi0/1.2002 602 up RX punts 26142Gi0/1.2004 604 upThe following example displays the interface information for VLAN number 100 on Gigabit Ethernet interface 0/1, including up state, receiving packet count, receiving bytes count, transmitting packet count, transmitting byte count, and receiving punt count:
Router# show mpf interface GigabitEthernet0/1 100Name Index State Counter CountGi0/1.100 100 up RX packets 1004RX bytes 158632TX packets 5004TX bytes 790632RX punts 25Table 12 describes the fields shown in the output examples.
Related Commands
show mpf punt
To display the Multi-Processor Forwarding (MPF) punt reason and punt packet count for the chassis, use the show mpf punt command in user EXEC or privileged EXEC mode.
show mpf punt
Syntax Description
This command has no arguments or keywords.
Command Modes
User EXEC
Privileged EXECCommand History
Usage Guidelines
The punt reason and punt packet count are collected for each box or chassis, not for each interface. Packets that are punted are directed for Cisco IOS processing and are not accelerated by MPF.
Examples
The following example displays the types of packet, the reasons for the punt, and the punt packet counts for the router chassis.
Router# show mpf puntType Message Countl2tp unknown session errors 7l2tp L2TP control 6ipv4/verify adjacency punt 1ethernet unknown ethernet type 542ppp punts due to unknown protocol 333arp ARP request 6Table 13 describes the fields in the show mpf punt output display.
Related Commands
sw-module heap fp
To fine-tune the Multi-Processor Forwarding (MPF) heap memory allocation required for specific session scaling and application needs, use the sw-module heap fp command in global configuration mode. To return the setting to the default (32 MB), use the no form of the command.
sw-module heap fp [megabytes]
no sw-module heap fp
Syntax Description
megabytes
(Optional) The heap size in megabytes (MB) for the MPF processor. The default size is 32 MB.
Command Default
The default heap memory allocation size is 32 MB.
Command Modes
Global configuration
Command History
Usage Guidelines
The default heap size is 32 MB if you do not specify otherwise. Once you have changed and saved the MPF heap memory configuration, reboot the router for the MPF memory size adjustment to take effect.
The following table lists the recommended heap memory size by type of deployment and number of sessions configured:
Table 14 Recommended Heap Memory Sizes
PTA/LAC/LNS
8000 and over
80 MB
Examples
The following example sets or changes the MPF heap memory size in a router to 80 MB:
Router(config)# sw-module heap fp 80Related Commands
Glossary
broadband—Transmission methodology that multiplexes multiple independent signals onto one cable. In telecommunications, broadband is classified as any channel with bandwidth greater than 4kHz (typical voice grade). In LAN terminology, broadband is classified as a coaxial cable on which analog signaling is employed.
FF—fast-forwarding. Accelerates broadband features by enabling the second CPU on the Cisco 7301 and Cisco 7200 VXR routers.
MPF accelerated features—A subset of Cisco IOS features that is fast forwarded to CPU1 for processing, which significantly improves performance by up to two times that of a regular Cisco 7301 or Cisco 7200 VXR router without any hardware changes.
All control plane and data plane traffic is fast forwarded to CPU1 for processing.
punted—Non-MPF features that are not accelerated; packets are punted to CPU0 where they are processed by standard Cisco IOS software running on CPU0.
ROMmon—ROM monitor. ROMmon is the initializing software that is executed when the router is powered up.
Copyright © 2005 Cisco Systems, Inc. All rights reserved.
