Saturday, October 20, 2007

Cisco IOS IP Accounting Features

Cisco IOS IP Accounting Features

Cisco IOS IP Accounting Features

Date: Oct 19, 2007 By Benoit Claise, Ralf Wolter. Sample Chapter is provided courtesy of Cisco Press.

This chapter describes the IP Accounting features in Cisco IOS and enables you to distinguish the different IP Accounting functions and understand SNMP MIB details. This chapter also provides a command-line reference.

IP Accounting is a very useful accounting feature in Cisco IOS, but it’s not as well known as other features, such as NetFlow. The fact that Cisco has considered replacing IP Accounting by adding new features to NetFlow potentially turns IP Accounting into a corner case solution. However, compared to NetFlow, IP Accounting offers some advantages that make it an interesting feature to investigate: easy results retrieval via a MIB and limited resource consumption. Furthermore, access-list accounting currently cannot be solved with the NetFlow implementation. Note that NetFlow recently added the export of the MAC address as a new information element. Refer to coverage of NetFlow Layer 2 and the Security Monitoring Exports feature in Chapter 7, “NetFlow.”

IP Accounting comes in four variations:

  • Basic IP Accounting, which this book calls “IP Accounting (Layer 3)”
  • IP Accounting Access Control List (ACL)
  • IP Accounting MAC Address
  • IP Accounting Precedence

Note that Cisco documentation is not always consistent for the different IP Accounting features. Therefore, this book uses the command-line interface (CLI) commands as titles, except for “IP Accounting Access Control List,” where the related CLI command is ip accounting access-violations.

This chapter discusses in detail each flavor of IP Accounting, using a basic structure. First, the fundamentals are explained, followed by an overview of CLI operations, and then SNMP operations. It concludes by comparing the IP Accounting features to the questions raised in Chapter 2, “Data Collection Methodology”:

  • What to collect?
  • Where and how to collect?
  • How to configure?
  • Who is the user?
  • Potential scenarios.

IP Accounting (Layer 3)

IP Accounting (Layer 3) collects the number of bytes and packets processed by the network element on a source and destination IP address basis. Only transit traffic that enters and leaves the router is measured, and only on an outbound basis. Traffic generated by the router or traffic terminating in the router is not included in the accounting statistics. IP Accounting (Layer 3) collects individual IP address details, so it can be used to identify specific users for usage-based billing. To provide the operator with the opportunity of “snapshot” collections in the network, IP Accounting (Layer 3) maintains two accounting databases: an active database and a checkpoint database. The active collection process always updates the active database and therefore constantly increments the counters while packets pass the router. To get a snapshot of the traffic statistics, a CLI command or SNMP request can be executed to copy the current status from the active database to the checkpoint database. This copy request can be automated across the network to be executed at the same time, and a Network Management application can later retrieve the accounting details from the checkpoint database to present consistent accounting data to the operator. The checkpoint database offers a “frozen” snapshot of the complete network. Trying to achieve the same result by synchronously polling entire MIB tables across multiple network elements would introduce some inaccuracies, and hence no real “frozen” snapshots. The collected data can be used for performance and trending applications that require collections at regular intervals. The snapshot function is unique to IP Accounting.

IP Accounting (Layer 3) Principles

The principles of IP Accounting (Layer 3) can be summarized as follows:

  • IP Layer 3 outbound (egress) traffic is collected.
  • Only transit traffic that enters and leaves the router is collected; traffic that is generated by the router or terminated in the router is not included.
  • IP Accounting (Layer 3) also collects IPX traffic. In this case, IPX source and destination addresses are reported instead of IP addresses.
  • Egress MPLS core traffic collection is a new feature.
  • Active and checkpoint databases enable “snapshot” collections.
  • Collection data is accessible via CLI and SNMP; however, the initial configuration must be done via CLI. To retrieve the collection results via SNMP, you need to enable SNMP first. When configuring SNMP, distinguish between read-only access and read-write access. For more details about SNMP configuration, see Chapter 4, “SNMP and MIBs.”
  • The MIB contains only 32-bit SNMP counters.
Supported Devices and IOS Versions

The following list defines the devices and Cisco IOS Software releases that support IP Accounting (Layer 3):

  • IP Accounting (Layer 3) was introduced in IOS 10.0.
  • It is supported on all routers, including Route Switch Module (RSM) and Multilayer Service Feature Card (MSFC), except for the Cisco 12000. Note that IP Accounting cannot account for MLS-switched traffic on the Catalyst 6500/7600, so it collects only a subset of traffic on these platforms.
  • It is supported on all physical interfaces and logical subinterfaces.
  • IP Accounting (Layer 3) runs on the top of all switching paths, except for autonomous switching, silicon switching engine (SSE) switching, and distributed switching (dCEF) on the interface. On the Cisco 7500 router, IP Accounting (Layer 3) causes packets to be switched on the Route Switch Processor (RSP) instead of the Versatile Interface Processor (VIP), which can cause additional performance degradation.
CLI Operations

Notable commands for configuring, verifying, and troubleshooting IP Accounting (Layer 3) are as follows:

  • router(config-if)# ip accounting output-packets

    enables IP Accounting (Layer 3) for output traffic on the interface.

  • router(config)# ip accounting-list [ip address] [ip address mask]

    defines filters to control the hosts for which IP Accounting (Layer 3) information is kept. The filters are similar to an aggregation scheme and can be used to reduce the number of collected records. If filters are applied, details such as number of packets and bytes are kept only for the traffic that matches the filters, while all others are aggregated into “transit records.”

  • router(config)# ip accounting-transits count

    controls the number of transit records that are stored in the IP Accounting (Layer 3) database. Transit entries are those that do not match any of the filters specified by the global configuration command ip accounting-list. If no filters are defined, no transit entries are possible. The default number of transit records that are stored in the IP Accounting (Layer 3) database is 0.

    Note that the term “transit” in this case refers to packets that are not matched by the filter statements. In the IP Accounting (Layer 3) definition, “transit” refers to packets that traverse the router, compared to traffic that is generated at the router or destined for the router.

  • router(config)# ip accounting-threshold count

    sets the maximum number of accounting entries to be created. The accounting threshold defines the maximum number of entries (source and destination address pairs) that are accumulated. The default accounting threshold is 512 entries, which results in a maximum table size of 12,928 bytes. The threshold counter applies to both the active and checkpoint tables.

    The threshold value depends on the traffic mix, because different traffic types create different records for the source and destination address pairs. Whenever the table is full, the new entries (overflows) are not accounted. However, show ip accounting displays the overflows: “Accounting threshold exceeded for X packets and Y bytes.” Alternatively, these values are available in the MIB: actLostPkts (lost IP packets due to memory limitations) and actLostByts (total bytes of lost IP packets). You should monitor the overflows number, at least during the deployment phase, to find the right balance between the number of entries and memory consumption.

  • router# show ip accounting [checkpoint] output-packets

    displays the active accounting or checkpoint database.

  • router# clear ip accounting

    copies the content of the active database to the checkpoint database and clears the active database afterward.

  • router# clear ip accounting checkpoint

    clears the checkpoint database.

NOTE

The IP Accounting (Layer 3) and IP Accounting Access Control List entries share the same databases. Consequently, there is no explicit command to erase the IP Accounting (Layer 3) entries independently of the IP Accounting ACL entries.

SNMP Operations

The OLD-CISCO-IP-MIB has two tables:

  • lipAccountingTable, the active database
  • lipCkAccountingTable, the checkpoint database

The MIB variable actCheckPoint must be read first and then set to the same value that was read to copy the active database into the checkpoint database. After a successful SNMP set request, actCheckPoint is incremented by 1. Setting actCheckPoint is the equivalent of the clear ip accounting CLI command. A Network Management application can retrieve the MIB variable lipCkAccountingTable to analyze stable data in the checkpoint database. There is no SNMP variable to erase the content of the checkpoint database; however, setting actCheckPoint again flushes the checkpoint database and copies the content of the active database.

Details of the IP Accounting MIB (OLD-CISCO-IP-MIB) are as follows:

  • Active database—The lipAccountingTable table contains four relevant components:

    • — actSrc is the active database source.
    • — actDst is the active database destination.
    • — actPkts is the active database packets.
    • — actByts is the active database bytes.

    The table indexes are actSrc and actDst.

  • Checkpoint database—The lipCkAccountingTable table contains four relevant components:

    • — ckactSrc is the checkpoint database source.
    • — ckactdDst is the checkpoint database destination.
    • — ckactPkts is the checkpoint database packets.
    • — ckactByts is the checkpoint database bytes.

    The table indexes are ckactSrc and ckactDst.

  • actCheckPoint MIB variable

NOTE

The active and checkpoint MIB tables contain an ACL violations entry. Because it is relevant only to the IP Accounting Access Control List, it is not discussed in this section.

Examples (CLI and SNMP)

The following example provides a systematic introduction for configuring and monitoring IP Accounting (Layer 3) and displays the results for both CLI and SNMP.

Initial Configuration

Initially, both the active database (lipAccountingTable) and checkpoint database (lipCkAccountingTable) are empty, as shown from the router CLI and from the SNMP tables. 

router#show ip accounting output-packets
   Source          Destination               Packets            Bytes
   Accounting data age is 0
router#show ip accounting checkpoint output-packet
   Source          Destination               Packets            Bytes
   Accounting data age is 0

The router is accessed with SNMP2c (SNMP version 2c), the read community string is public, and the SNMP tool net-snmp is used.

SERVER % snmpwalk -c public -v 2c <router> lipAccountingTable
    actDst.0.0.0.0.0.0.0.0 = IpAddress: 0.0.0.0
    actByts.0.0.0.0.0.0.0.0 = INTEGER: 0
SERVER % snmpwalk -c public -v 2c <router> lipCkAccountingTable
    ckactDst.0.0.0.0.0.0.0.0 = IpAddress: 0.0.0.0
    ckactByts.0.0.0.0.0.0.0.0 = INTEGER: 0

The IP Accounting (Layer 3) configuration is straightforward:

router(config)#int serial 0/0
router(config-if)#ip accounting output-packets
router(config-if)#exit

Collection Monitoring

After configuring IP Accounting (Layer 3), the active database populates:

router#show ip accounting output-packet
   Source           Destination             Packets         Bytes
 192.1.1.110        192.1.1.97                 5             500
 192.1.1.110        192.1.1.26                 5             500

The corresponding MIB table shows the identical entries:

SERVER % snmptable -Ci -Cb -c public -v 2c <router> lipAccountingTable
                           index           Src          Dst   Pkts   Byts
          192.1.1.110.192.1.1.26 192.1.1.110 192.1.1.26          5    500
          192.1.1.110.192.1.1.97 192.1.1.110 192.1.1.97          5    500

At this point, the checkpoint database is still empty. The active database content is cleared by copying its content to the checkpoint database: 

router#clear ip accounting

As an alternative, the clear ip accounting mechanism can be mimicked by using the actCheckPoint MIB variable procedure. That means reading the content of the MIB variable and setting it again to the same value that was read: 

SERVER % snmpget -c public -v 2c <router> actCheckPoint.0
       actCheckPoint.0 = INTEGER: 0
SERVER % snmpset -c private -v 2c <router> actCheckPoint.0 i 0
       actCheckPoint.0 = INTEGER: 0
      SERVER % snmpget -c public -v 2c <router> actCheckPoint.0
       actCheckPoint.0 = INTEGER: 1

The two entries just discussed are now in the checkpoint database, but the active database is empty:

router#show ip accounting output-packets
   Source           Destination             Packets              Bytes
   Accounting data age is 0
router#show ip accounting output-packets checkpoint output-packets
   Source           Destination             Packets             Bytes
 192.1.1.110        192.1.1.97                 5                 500
 192.1.1.110        192.1.1.26                 5                 500
SERVER % snmptable -Ci -Cb -c public -v 2c <router> lipCkAccountingTable
                           index           Src          Dst   Pkts   Byts
          192.1.1.110.192.1.1.26 192.1.1.110 192.1.1.26          5    500
          192.1.1.110.192.1.1.97 192.1.1.110 192.1.1.97          5    500
Posted by at
Labels: