The PFC and DFC provide hardware support for classes that match multicast traffic.
CoPP is not supported in hardware for broadcast packets. The combination of ACLs, traffic storm control, and CoPP software protection provides protection against broadcast DoS attacks.
CoPP does not support non-IP classes except for the default non-IP class. ACLs can be used instead of non-IP classes to drop non-IP traffic, and the default non-IP CoPP class can be used to limit to non-IP traffic that reaches the RP CPU.
Do not use the
keyword in CoPP policy ACLs.
If you have a large QoS configuration, the system may run out of TCAM space. If this is the case, CoPP may be performed in software.
When there is a large QoS configuration for other interfaces, you can run out of TCAM space. When this situation occurs, CoPP may be performed entirely in software and result in performance degradation and CPU cycle consumption.
You must ensure that the CoPP policy does not filter critical traffic such as routing protocols or interactive access to the switches. Filtering this traffic could prevent remote access to the switch, requiring a console connection.
The PFC and DFCs support built-in special-case rate limiters, which are useful for situations where an ACL cannot be used (for example, TTL, MTU, and IP options). When you enable the special-case rate limiters, you should be aware that the special-case rate limiters will override the CoPP policy for packets matching the rate-limiter criteria.
Neither egress CoPP nor silent mode is supported. CoPP is only supported on ingress (service-policy output CoPP cannot be applied to the control plane interface).
ACE hit counters in hardware are only for ACL logic. You can rely on software ACE hit counters and the show access-list, show policy-map control-plane, and show platform ip qos commands to troubleshoot evaluate CPU traffic.
CoPP is performed on a per-forwarding-engine basis and software CoPP is performed on an aggregate basis.
CoPP does not support ACEs with the
CoPP uses hardware QoS TCAM resources. Enter the
show tcam utilization
command to verify the TCAM utilization.
To avoid matching the filtering and policing that are configured in a subsequent class, configure policing in each class. CoPP does not apply the filtering in a class that does not contain a police command. A class without a police command matches no traffic.
The ACLs used for classification are QoS ACLs. The supported QoS ACLs are IP standard, extended, and named.
These are the only match types supported:
Only IP ACLs are supported in hardware.
MAC-based matching is done in software only.
You can enter one
command in a single class map only.
When defining the service policy, the
policy-map action is the only supported action.
When applying the service policy to the control plane, the
direction is only supported.
Information About CoPP
The traffic managed by the RP is divided into three functional components or
The control plane policing (CoPP) feature increases security on the switch by protecting the RP from unnecessary or DoS traffic and giving priority to important control plane and management traffic. The PFC and DFCs provide hardware support for CoPP. CoPP works with the hardware rate limiters.
The PFC and DFCs support the built-in “special case” rate limiters that can be used when an ACL cannot classify particular scenarios, such as IP options cases, TTL and MTU failure cases, packets with errors, and multicast packets. When enabling the special-case rate limiters, the special-case rate limiters override the CoPP policy for packets matching the rate-limiter criteria.
The majority of traffic managed by the RP is handled by way of the control and management planes. You can use CoPP to protect the control and management planes, and ensure routing stability, reachability, and packet delivery. CoPP uses a dedicated control plane configuration through the modular QoS CLI (MQC) to provide filtering and rate-limiting capabilities for the control plane packets.
You must first identify the traffic to be classified by defining a class map. The class map defines packets for a particular traffic class. After you have classified the traffic, you can create policy maps to enforce policy actions for the identified traffic.
Default Settings for CoPP
CoPP is enabled by default. To disable the default CoPP configuration, enter the
no service-policy input policy-default-autocopp
control plane configuration mode command.
You can define any number of classes, but typically traffic is grouped into classes that are based on relative importance. The following provides a sample grouping:
Border Gateway Protocol (BGP)—Traffic that is crucial to maintaining neighbor relationships for BGP routing protocol, for example, BGP keepalives and routing updates. Maintaining BGP routing protocol is crucial to maintaining connectivity within a network or to a service provider. Sites that do not run BGP do not need to use this class.
Interior Gateway Protocol (IGP)—Traffic that is crucial to maintaining IGP routing protocols, for example, open shortest path first OSPF, enhanced interior gateway routing protocol (EIGRP), and routing information protocol (RIP). Maintaining IGP routing protocols is crucial to maintaining connectivity within a network.
Management—Necessary, frequently used traffic that is required during day-to-day operations. For example, traffic used for remote network access, and Cisco IOS image upgrades and management, such as Telnet, secure shell (SSH), network time protocol (NTP), simple network management protocol (SNMP), terminal access controller access control system (TACACS), hypertext transfer protocol (HTTP), trivial file transfer protocol (TFTP), and file transfer protocol (FTP).
Reporting—Traffic used for generating network performance statistics for the purpose of reporting. For example, using Cisco IOS IP service level agreements (SLAs) to generate ICMP with different DSCP settings in order to report on response times within different QoS data classes.
Monitoring—Traffic used for monitoring a switch. Traffic should be permitted but should never be a risk to the switch; with CoPP, this traffic can be permitted but limited to a low rate. For example, ICMP echo request (ping) and traceroute.
Critical Applications—Critical application traffic that is specific and crucial to a particular customer environment. Traffic included in this class should be tailored specifically to the required application requirements of the user (in other words, one customer may use multicast, while another uses IPsec or generic routing encapsulation (GRE). For example, GRE, hot standby router protocol (HSRP), virtual router redundancy protocol (VRRP), session initiation protocol (SIP), data link switching (DLSw), dynamic host configuration protocol (DHCP), multicast source discovery protocol (MSDP), Internet group management protocol (IGMP), protocol independent multicast (PIM), multicast traffic, and IPsec.
Layer 2 Protocols—Traffic used for address resolution protocol (ARP). Excessive ARP packets can potentially monopolize RP resources, starving other important processes; CoPP can be used to rate limit ARP packets to prevent this situation. ARP is the only Layer 2 protocol that can be specifically classified using the match protocol classification criteria.
Undesirable—Explicitly identifies bad or malicious traffic that should be unconditionally dropped and denied access to the RP. The undesirable classification is particularly useful when known traffic destined for the switch should always be denied and not placed into a default category. If you explicitly deny traffic, then you can enter
commands to collect approximate statistics on the denied traffic and estimate its rate.
Default—All remaining traffic destined for the RP that has not been identified. MQC provides the default class, so the user can specify the treatment to be applied to traffic not explicitly identified in the other user-defined classes. This traffic has a highly reduced rate of access to the RP. With a default classification in place, statistics can be monitored to determine the rate of otherwise unidentified traffic destined for the control plane. After this traffic is identified, further analysis can be performed to classify it and, if needed, the other CoPP policy entries can be updated to accomodate this traffic.
Before you develop the actual CoPP policy, you must identify and separate the required traffic into different classes. Traffic is grouped into nine classes that are based on relative importance. The actual number of classes needed might differ and should be selected based on your local requirements and security policies.
You do not have to define policies that match bidirectionally. You only need to identify traffic unidirectionally (from the network to the RP) since the policy is applied on ingress only.
Sample Basic ACLs for CoPP Traffic Classification
This section shows sample basic ACLs for CoPP classification. In the samples, the commonly required traffic is identified with these ACLs:
ACL 120—Critical traffic
ACL 121—Important traffic
ACL 122—Normal traffic
ACL 123—Explicitly denies unwanted traffic
ACL 124—All other traffic
This example shows how to define ACL 120 for critical traffic:
Router(config)# access-list 120 remark CoPP ACL for critical traffic
This example shows how to allow BGP from a known peer to this switch’s BGP TCP port:
This example shows how to define ACL 122 for the normal traffic class:
Router(config)# access-list 122 remark CoPP normal traffic
This example shows how to permit switch-originated traceroute traffic:
Router(config)# access-list 122 permit icmp any any ttl-exceeded
Router(config)# access-list 122 permit icmp any any port-unreachable
This example shows how to permit receipt of responses to the switch that originated the pings:
Router(config)# access-list 122 permit icmp any any echo-reply
This example shows how to allow pings to the switch:
Router(config)# access-list 122 permit icmp any any echo
This example shows how to define ACL 123 for the undesirable class.
Router(config)# access-list 123 remark explicitly defined "undesirable" traffic
Note In the following example, ACL 123 is a permit entry for classification and monitoring purposes, and traffic is dropped as a result of the CoPP policy.
This example shows how to permit all traffic destined to UDP 1434 for policing:
Router(config)# access-list 123 permit udp any any eq 1434
This example shows how to define ACL 124 for all other traffic:
Router(config)# access-list 124 remark rest of the IP traffic for CoPP
Router(config)# access-list 124 permit ip any any
You can enter the
show policy-map control-plane
command for developing site-specific policies, monitoring statistics for the control plane policy, and troubleshooting CoPP. This command displays dynamic information about the actual policy applied, including rate information and the number of bytes (and packets) that conformed or exceeded the configured policies both in hardware and in software.
The output of the
show policy-map control-plane
command is as follows: