Catalyst Supervisor Engine 32 PISA IOS Software Configuration Guide, 12.2ZY

Understanding Cisco IOS ACL Support

This chapter describes Cisco IOS ACL support on the Catalyst 6500 series switches:

•Cisco IOS ACL Configuration Guidelines and Restrictions

•Hardware and Software ACL Support

•Optimized ACL Logging with a PFC3B

•Guidelines and Restrictions for Using Layer 4 Operators in ACLs

Note For complete information about configuring Cisco IOS ACLs, see the Cisco IOS Security Configuration Guide, Release 12.2, "Traffic Filtering and Firewalls," at this URL:

http://www.cisco.com/en/US/docs/ios-xml/ios/sec_data_acl/configuration/12-2sx/sec-data-acl-12-2sx-book.html

Cisco IOS ACL Configuration Guidelines and Restrictions

The following guidelines and restrictions apply to Cisco IOS ACL configurations:

•You can apply Cisco IOS ACLs directly to Layer 3 ports and to VLAN interfaces.

•You can apply VLAN ACLs (VACLs) to VLANs (refer to Chapter 32 "Configuring VLAN ACLs").

•Each type of ACL (IP, IPX, and MAC) filters only traffic of the corresponding type. A Cisco IOS MAC ACL never matches IP or IPX traffic.

•The PFC3B does not provide hardware support for Cisco IOS IPX ACLs. Cisco IOS IPX ACLs are supported in software on the PISA.

•By default, the PISA sends Internet Control Message Protocol (ICMP) unreachable messages when a packet is denied by an access group.

With the ip unreachables command enabled (which is the default), the supervisor engine drops most of the denied packets in hardware and sends only a small number of packets to the PISA to be dropped (10 packets per second, maximum), which generates ICMP-unreachable messages.

To eliminate the load imposed on the PISA CPU by the task of dropping denied packets and generating ICMP-unreachable messages, you can enter the no ip unreachables interface configuration command to disable ICMP unreachable messages, which allows all access group-denied packets to be dropped in hardware.

•ICMP unreachable messages are not sent if a packet is denied by a VACL.

•We strongly recommend that you use named ACLs (rather than numbered ACLs) as this conserves CPU usage when creating or modifying ACL configurations and during system restarts. When you create ACL entries (or modify existing ACL entries), the software performs a CPU-intensive operation called an ACL merge to load the ACL configurations into the PFC hardware. An ACL merge also occurs when the startup configuration is applied during a system restart.

With named ACLs, the ACL Merge is triggered only when the user exits the named-acl configuration mode. However with numbered ACLs, the ACL Merge is triggered for every ACE definition and results in a number of intermediate merges during ACL configuration.

Hardware and Software ACL Support

Access control lists (ACLs) can be processed in hardware by the PFC3B or in software by the PISA. The following behavior describes software and hardware handling of ACLs:

•The PFC3B provides more efficient hardware support for named ACLs than it can for numbered ACLs.

•ACL flows that match a "deny" statement in standard and extended ACLs (input and output) are dropped in hardware if "ip unreachables" is disabled.

•ACL flows that match a "permit" statement in standard and extended ACLs (input and output) are processed in hardware.

•VLAN ACL (VACL) flows are processed in hardware. If a field specified in a VACL is not supported by hardware processing that field is ignored (for example, the log keyword in an ACL) or the whole configuration is rejected (for example, a VACL containing IPX ACL parameters).

•VACL logging is processed in software.

•Dynamic ACL flows are processed in hardware.

•Idle timeout is processed in software.

Note Idle timeout is not configurable. Catalyst 6500 series switches do not support the access-enable host timeout command.

•Except on MPLS interfaces, reflexive ACL flows are processed in hardware after the first packet in a session is processed in software on the RP.

•IP accounting for an ACL access violation on a given port is supported by forwarding all denied packets for that port to the PISA for software processing without impacting other flows.

•The PFC3B does not provide hardware support for Cisco IOS IPX ACLs. Cisco IOS IPX ACLs are supported in software on the PISA.

•Extended name-based MAC address ACLs are supported in hardware.

•The following ACL types are processed in software:

–Internetwork Packet Exchange (IPX) access lists

–Standard XNS access list

–Extended XNS access list

–DECnet access list

–Extended MAC address access list

–Protocol type-code access list

Note IP packets with a header length of less than five will not be access controlled.

•Unless you configure optimized ACL logging (OAL), flows that require logging are processed in software without impacting nonlogged flow processing in hardware (see the "Optimized ACL Logging with a PFC3B" section).

•The forwarding rate for software-processed flows is substantially less than for hardware-processed flows.

•When you enter the show ip access-list command, the match count displayed does not include packets processed in hardware.

•When you enter the show policy-map interface command, sometimes the counters that are displayed do not include all of the hardware switching platform counters.

Optimized ACL Logging with a PFC3B

These sections describe optimized ACL logging (OAL):

•Understanding OAL

•OAL Guidelines and Restrictions

•Configuring OAL

Understanding OAL

OAL provides hardware support for ACL logging. Unless you configure OAL, packets that require logging are processed completely in software on the PISA. OAL permits or drops packets in hardware on the PFC3B and uses an optimized routine to send information to the PISA to generate the logging messages.

OAL Guidelines and Restrictions

The following guidelines and restrictions apply to OAL:

•OAL and VACL capture are incompatible. Do not configure both features on the switch. With OAL configured, use SPAN to capture traffic.

•OAL is supported only on the PFC3B.

•OAL supports only IPv4 unicast packets.

•OAL supports VACL logging of permitted ingress traffic.

•OAL does not support port ACLs (PACLs).

•OAL does not provide hardware support for the following:

–Reflexive ACLs

–ACLs used to filter traffic for other features (for example, QoS)

–Exception packets (for example, TTL failure and MTU failure)

–Packets with IP options

–Packets addressed at Layer 3 to the router

–Packets sent to the PISA to generate ICMP unreachable messages

–Packets being processed by features not accelerated in hardware

•To provide OAL support for denied packets, enter the mls rate-limit unicast ip icmp unreachable acl-drop 0 command.

•OAL and the mls verify ip length minimum command are incompatible. Do not configure both.

Configuring OAL

These sections describe how to configure OAL:

•Configuring OAL Global Parameters

•Configuring OAL on an Interface

•Displaying OAL Information

•Clearing Cached OAL Entries

Note•For complete syntax and usage information for the commands used in this section, refer to the Catalyst Supervisor Engine 32 PISA Cisco IOS Command Reference, Release 12.2ZY.

•To provide OAL support for denied packets, enter the mls rate-limit unicast ip icmp unreachable acl-drop 0 command.

Configuring OAL Global Parameters

To configure global OAL parameters, perform this task:

When configuring OAL global parameters, note the following information:

•entries number_of_entries

–Sets the maximum number of entries cached.

–Range: 0-1,048,576 (entered without commas).

–Default: 8192.

•interval seconds

–Sets the maximum time interval before an entry is sent to be logged. Also if the entry is inactive for this duration it is removed from the cache.

–Range: 5-86,400 (1440 minutes or 24 hours, entered without commas).

–Default: 300 seconds (5 minutes).

•rate-limit number_of_packets

–Sets the number of packets logged per second in software.

–Range: 10-1,000,000 (entered without commas).

–Default: 0 (rate limiting is off and all packets are logged).

•threshold number_of_packets

–Sets the number of packet matches before an entry is logged.

–Range: 1-1,000,000 (entered without commas).

–Default: 0 (logging is not triggered by the number of packet matches).

Configuring OAL on an Interface

To configure OAL on an interface, perform this task:

Displaying OAL Information

To display OAL information, perform this task:

Router # show logging ip access-list cache 

Clearing Cached OAL Entries

To clear cached OAL entries, perform this task:

Router # clear logging ip access-list cache 

Guidelines and Restrictions for Using Layer 4 Operators in ACLs

These sections describe guidelines and restrictions when configuring ACLs that include Layer 4 port operations:

•Determining Layer 4 Operation Usage

•Determining Logical Operation Unit Usage

Determining Layer 4 Operation Usage

You can specify these types of operations:

•gt (greater than)

•lt (less than)

•neq (not equal)

•eq (equal)

•range (inclusive range)

We recommend that you do not specify more than nine different operations on the same ACL. If you exceed this number, each new operation might cause the affected ACE to be translated into more than one ACE.

Use the following two guidelines to determine Layer 4 operation usage:

•Layer 4 operations are considered different if the operator or the operand differ. For example, in this ACL there are three different Layer 4 operations ("gt 10" and "gt 11" are considered two different Layer 4 operations):

... gt 10 permit

... lt 9 deny

... gt 11 deny

Note There is no limit to the use of "eq" operators as the "eq" operator does not use a logical operator unit (LOU) or a Layer 4 operation bit. See the "Determining Logical Operation Unit Usage" section for a description of LOUs.

•Layer 4 operations are considered different if the same operator/operand couple applies once to a source port and once to a destination port. For example, in this ACL there are two different Layer 4 operations because one ACE applies to the source port and one applies to the destination port.

... Src gt 10 ...

... Dst gt 10

Determining Logical Operation Unit Usage

Logical operation units (LOUs) are registers that store operator-operand couples. All ACLs use LOUs. There can be up to 32 LOUs; each LOU can store two different operator-operand couples with the exception of the range operator. LOU usage per Layer 4 operation is as follows:

•gt uses 1/2 LOU

•lt uses 1/2 LOU

•neq uses 1/2 LOU

•range uses 1 LOU

•eq does not require a LOU

For example, this ACL would use a single LOU to store two different operator-operand couples:

... Src gt 10 ...

... Dst gt 10

A more detailed example follows:

ACL1

... (dst port) gt 10 permit

... (dst port) lt 9 deny

... (dst port) gt 11 deny

... (dst port) neq 6 permit

... (src port) neq 6 deny

... (dst port) gt 10 deny

ACL2

... (dst port) gt 20 deny

... (src port) lt 9 deny

... (src port) range 11 13 deny

... (dst port) neq 6 permit

The Layer 4 operations and LOU usage is as follows:

•ACL1 Layer 4 operations: 5

•ACL2 Layer 4 operations: 4

•LOUs: 4

An explanation of the LOU usage follows:

•LOU 1 stores "gt 10" and "lt 9"

•LOU 2 stores "gt 11" and "neq 6"

•LOU 3 stores "gt 20" (with space for one more)

•LOU 4 stores "range 11 13" (range needs the entire LOU)