Catalyst 4500 Series Switch Cisco IOS Software Configuration Guide, 12.2(20)EWA

Configuring NetFlow Statistics Collection

This chapter describes how to configure NetFlow statistics on the Catalyst 4500 series switches. It also provides guidelines, procedures, and configuration examples.

Note This feature is only available if the NetFlow Services Card (WS-F4531) is present.

Note For complete syntax and usage information for the switch commands used in this chapter, first look at the Cisco Catalyst 4500 Series Switch Command Reference and related publications at this location:

http://www.cisco.com/en/US/products//hw/switches/ps4324/index.html

If the command is not found in the Catalyst 4500 Command Reference, it will be found in the larger Cisco IOS library. Refer to the Catalyst 4500 Series Switch Cisco IOS Command Reference and related publications at this location:

http://www.cisco.com/en/US/products/ps6350/index.html

The following topics are included:

•Overview of NetFlow Statistics Collection

•Caveat for the NetFlow Feature

•Configuring NetFlow Statistics Collection

•Configuring NetFlow Aging Parameters

•NetFlow Statistics Collection Configuration Example

•NetFlow Configuration Examples

Overview of NetFlow Statistics Collection

This section contains the following subsections:

•Information Derived from Hardware

•Information Derived from Software

•Determining the Input and Output interface and AS Numbers

•VLAN Statistics

A network flow is defined as a unidirectional sequence of packets between a given source and destination endpoints. Network flows are highly granular; flow endpoints are identified both by an IP address and transport layer application port numbers. NetFlow also utilizes the IP type and the input interface identifier to uniquely identify flows.

NetFlow statistics is a global traffic monitoring feature that allows flow-level monitoring of all IPv4-routed traffic through the switch. Collected statistics can be exported to an external device (NetFlow Collector/Analyzer) through the NetFlow Data Export (NDE). Network planners can selectively enable NetFlow statistics (and NDE) on a per-device basis to gain traffic performance, control, or accounting benefits in specific network locations. Traffic monitoring does not need to be operating on each device in the network.

Information Derived from Hardware

Information available in a typical NetFlow record from hardware includes the following:

•the packet and byte counts

•start and end timestamps

•source and destination IP addresses

•IP protocol

•source and destination port numbers.

Information Derived from Software

The software infers the following fields:

•Input identifier

•Output identifier

•Routing information, including next-hop address, origin and peer AS, source and destination prefix mask

Determining the Input and Output interface and AS Numbers

The input and output interface values determined by the software, however, are not guaranteed to be accurate in the specific situations explained below. The input and output interface values are correct when Policy Based Routing (PBR) is not used. There are simple symmetric routing scenarios, and no load balancing schemes are applied by an adjacent upstream router.

Software determines the output interface information by looking up the FIB (Forwarding Information Base) entry in the default FIB table (based on the destination IP address). From this FIB entry, the software gains access to the destination AS number for this destination IP address, as well as the appropriate adjacency that stores the interface information. Therefore, the output interface is based solely on the destination IP address. If load balancing is enabled on the switch, instead of looking at the adjacency in the FIB entry, the load balancing hash will be applied to access the appropriate FIB path and access the appropriate adjacency. Although this process will typically yield correct results, a potential inaccuracy can occur when using a PBR that shares IP addresses with the default FIB table. Under these circumstances, there would then be multiple FIB table entries and associated adjacencies for the same destination IP address.

Similarly, the input interface and the source AS number for the source IP address are determined by looking up the FIB entry in the default FIB table based on the source IP address. Therefore, the input interface is based solely on the source IP address and a reverse lookup is done to determine to which interface a packet with this IP destination address needs to be routed. This process assumes that the forwarding paths are symmetrical. However, if this process yields multiple input interfaces, a deterministic algorithm will be applied to pick one of them—the one with the lowest IP address. Although this process typically yields correct values, there are scenarios where the values are inaccurate:

•If load balancing is being applied by an upstream adjacent router, one input interface must be chosen arbitrarily out of the multiple input interfaces available. This action is necessary because the input interface that would be used depends on the type of load balancing algorithm being deployed by the adjacent upstream router. It is not always feasible to know the algorithm. Therefore, all flow statistics will be attributed to one input interface. Software selects the interface with the lowest IP subnet number.

•In an asymmetric routing scheme, where the traffic for an IP subnet might be received on an interface that is different from the interface where packets are sent to this IP subnet, the inferences noted previously for selecting an input interface, based on a reverse lookup, would be incorrect and cannot be verified. The reason is that RPF-based forwarding is not supported in the Catalyst 4500 series switch Cisco IOS-based supervisor engines.

•If PBR (Policy Based Routing) is enabled on the switch and the flow is destined to an address that resides in the PBR range or is sourced from an address that resides in the PBR range, the information will be incorrect. In this case, the input and output interface will most likely point to the default route (if configured) or will have no value at all (NULL).

•In the case of detecting DOS attacks, even if the route is symmetrical, it might not yield the correct results because there may be multiple paths to the destination, which is the source of the attacks.

VLAN Statistics

The NetFlow Services module, in combination with the Catalyst 4500 series switch Supervisor Engine IV, provides the capability of reporting VLAN statistics for routed traffic in and out of a VLAN, as well as Layer 2 output VLAN statistics. The CLI output for a specific VLAN is shown below:

cat4k-sup4-2# sh vlan counters or show vlan id 22 count

* Multicast counters include broadcast packets

Vlan Id                                            :22

L2 Unicast Packets                                 :38

L2 Unicast Octets                                  :2432

L3 Input Unicast Packets                           :14344621

L3 Input Unicast Octets                            :659852566

L3 Output Unicast Packets                          :8983050

L3 Output Unicast Octets                           :413220300

L3 Output Multicast Packets                        :0

L3 Output Multicast Octets                         :0

L3 Input Multicast Packets                         :0

L3 Input Multicast Octets                          :0

L2 Multicast Packets                               :340

L2 Multicast Octets                                :21760

Caveat for the NetFlow Feature

The NetFlow Services module has hardware limitations that restrict the platform support to a subset of all NetFlow fields. Specifically, the following fields will not be supported:

•TCP Flags

•ToS

Configuring NetFlow Statistics Collection

To configure NetFlow Switching, complete the tasks in these sections:

•Checking for Required Hardware

•Enabling NetFlow Statistics Collection

•Exporting NetFlow Statistics

•Managing NetFlow Statistics Collection

•Configuring an Aggregation Cache

•Configuring a NetFlow Minimum Prefix Mask for Router-Based Aggregation

Checking for Required Hardware

Note The Catalyst 4500 series switches require WS-F4531 daughter cards and the Catalyst 4500 series switch Supervisor Engine IV to support the NetFlow statistics collection.

To ensure that the necessary hardware is enabled, issue the show module command, as follows:

Switch# show module all

Chassis Type : WS-C4507R

Power consumed by backplane : 40 Watts

Mod Ports Card Type                              Model              

Serial No.

---+-----+--------------------------------------+------------------+-----------

1     2  1000BaseX (GBIC) Supervisor(active)    WS-X4515           

JAB062604KB

2     2  1000BaseX (GBIC) Supervisor(standby)   WS-X4515           

JAB062408CB

6    48  10/100BaseTX (RJ45)                    WS-X4148           

JAB032305UH

M MAC addresses                    Hw  Fw           Sw               Status

--+--------------------------------+---+------------+----------------+---------

1 0001.6442.2c00 to 0001.6442.2c01 0.4 12.1(14r)EW( 12.1(20030513:00 Ok       

2 0001.6442.2c02 to 0001.6442.2c03 0.4 12.1(14r)EW( 12.1(20030513:00 Ok       

6 0050.3ed8.6780 to 0050.3ed8.67af 1.6 12.1(14r)EW( 12.1(20030513:00 Ok

Mod  Submodule               Model             Serial No.   Hw   Status

----+-----------------------+-----------------+------------+----+---------

1   Netflow Services Card   WS-F4531          JAB062209CG  0.2  Ok       

2   Netflow Services Card   WS-F4531          JAB062209AG  0.2  Ok       

Switch#

Note Enabling this feature does not impact the hardware-forwarding performance of the switch.

The effective size of the hardware flow "cache" table is 64 kilobytes. If more than 64 kilobytes flows are active simultaneously, statistics may be lost for some of the flows.

The effective size of the software flow table is 256 kilobytes. The NetFlow software manages the consistency between the hardware and software tables, keeping the hardware table open by purging inactive hardware flows to the software table.

User-configured timeout settings dictate when the flows are purged and exported through NDE from the software cache. Hardware flow management ensures consistency between hardware flow purging and the user-configured timeout settings.

Software-forwarded flows are also monitored. Moreover, statistics will overflow if any flow receives traffic at a sustained rate exceeding 2 gigabits per second. Generally, this situation should not occur because a port cannot transmit at a rate higher than 1 gigabit per second.

Note By design, even if the timeout settings are high, flows will automatically "age out" as they approach their statistics limit.

Enabling NetFlow Statistics Collection

Note Netflow Flow Statistics are disabled by default.

To enable NetFlow switching, first configure the switch for IP routing as described in the IP configuration chapters in the Cisco IOS IP and IP Routing Configuration Guide. After you configure IP routing, perform one of these tasks:

Exporting NetFlow Statistics

To configure the switch to export NetFlow statistics to a workstation when a flow expires, perform one of these tasks:

Managing NetFlow Statistics Collection

You can display and clear NetFlow statistics, including IP flow switching cache information and flow information, such as the protocol, total flow, flows per second, and so forth. You can also use the resulting information to obtain information about your switch traffic.

To manage NetFlow switching statistics, perform one or both of following tasks:

Configuring an Aggregation Cache

Aggregation of NetFlow statistics is typically performed by NetFlow collection tools on management workstations. By extending this support to the Catalyst 4500 series switch, you are now able to do the following:

•Reduce the required bandwidth between the router and workstations, as fewer NDE packets are exported.

•Reduce the number of collection workstations required.

•Provide visibility to aggregated flow statistics at the CLI.

To configure an aggregation cache, you must enter the aggregation cache configuration mode, and you must decide which type of aggregation scheme you would like to configure: autonomous system, destination prefix, protocol prefix, or source prefix aggregation cache. Once you define the aggregation scheme, define the operational parameters for that scheme. More than one aggregation cache can be configured concurrently.

To configure an aggregation cache, perform this task:

Verifying Aggregation Cache Configuration and Data Export

To verify the aggregation cache information, perform this task:

To confirm data export, perform the following task:

Configuring a NetFlow Minimum Prefix Mask for Router-Based Aggregation

The minimum prefix mask specifies the shortest subnet mask that will be used for aggregating flows within one of the IP-address based aggregation caches (e.g. source-prefix, destination-prefix, prefix). In these caches, flows are aggregated based upon the IP address (source, destination, or both, respectively) and masked by the longer of the Minimum Prefix mask and the subnet mask of the route to the source/destination host of the flow (as found in the switch routing table).

Note The default value of the minimum mask is zero. The configurable range for the minimum mask is from 1 to 32. You should chose an appropriate value depending on the traffic. A higher value for the minimum mask will provide more detailed network addresses, but it may also result in increased number of flows in the aggregation cache.

To configure a minimum prefix mask for the Router-Based Aggregation feature, perform the tasks described in the following sections. Each task is optional.

•Configuring the Minimum Mask of a Prefix Aggregation Scheme

•Configuring the Minimum Mask of a Destination-Prefix Aggregation Scheme

•Configuring the Minimum Mask of a Source-Prefix Aggregation Scheme

•Monitoring and Maintaining Minimum Masks for Aggregation Schemes

Configuring the Minimum Mask of a Prefix Aggregation Scheme

To configure the minimum mask of a prefix aggregation scheme, perform this task:

Configuring the Minimum Mask of a Destination-Prefix Aggregation Scheme

To configure the minimum mask of a destination-prefix aggregation scheme, perform this task:

Configuring the Minimum Mask of a Source-Prefix Aggregation Scheme

To configure the minimum mask of a source-prefix aggregation scheme, perform this task:

Monitoring and Maintaining Minimum Masks for Aggregation Schemes

To view the configured value of the minimum mask, use the following commands for each aggregation scheme, as needed:

Configuring NetFlow Aging Parameters

You can control when flows are purged from the software flow cache (and, if configured, reported through NDE) with the configuration aging parameters, Active and Inactive, of the ip flow-cache timeout command.

Active Aging specifies the period of time in which a flow should be removed from the software flow cache after the flow is created. Generally, this parameter is used to periodically notify external collection devices about active flows. This parameter operates independently of existing traffic on the flow. Active timeout settings tend to be on the order of minutes (default is 30min).

Inactive Aging specifies how long after the last packet is seen a flow is removed. The Inactive parameter clears the flow cache of "stale" flows thereby preventing new flows from starving (due to lack of resources). Inactive timeout settings tend to be on the order of seconds (default is 15sec).

NetFlow Statistics Collection Configuration Example

The following example shows how to modify the configuration to enable NetFlow switching. It also shows how to export the flow statistics for further processing to UDP port 9991 on a workstation with the IP address of 40.0.0.2. In this example, existing NetFlow statistics are cleared, thereby ensuring that the show ip cache flow command displays an accurate summary of the NetFlow switching statistics:

Switch# config t

Enter configuration commands, one per line.  End with CNTL/Z.

Switch(config)# ip route-cache flow

Switch(config)# ip flow-export destination 40.0.0.2 9991

Switch(config)# ip flow-export version 5

Switch(config)# end

Switch# show ip flow export

Flow export is enabled

  Exporting flows to 40.0.0.2 (9991)

  Exporting using source IP address 40.0.0.1

  Version 5 flow records

  2 flows exported in 1 udp datagrams

  0 flows failed due to lack of export packet

  0 export packets were sent up to process level

  0 export packets were dropped due to no fib

  0 export packets were dropped due to adjacency issues

  0 export packets were dropped due to fragmentation failures

  0 export packets were dropped due to encapsulation fixup failures

Switch#

Switch# show ip cache flow

IP Flow Switching Cache, 17826816 bytes

  0 active, 262144 inactive, 4 added

  14 ager polls, 0 flow alloc failures

  Active flows timeout in 1 minutes

  Inactive flows timeout in 10 seconds

  last clearing of statistics 15:48:37

Protocol         Total    Flows   Packets Bytes  Packets Active(Sec) Idle(Sec)

--------         Flows     /Sec     /Flow  /Pkt     /Sec     /Flow     /Flow

UDP-other            1      0.0         3    46      0.0       0.0      10.3

IP-other             1      0.0       100    38      0.0       0.0      10.2

Total:              2      0.0        51    38      0.0       0.0      10.2

SrcIf         SrcIPaddress    DstIf         DstIPaddress    Pr SrcP DstP  Pkts

Switch#

NetFlow Configuration Examples

This section provides the following basic configuration examples:

•Sample NetFlow Enabling Schemes

•Sample NetFlow Aggregation Configurations

•Sample NetFlow Minimum Prefix Mask Router-Based Aggregation Schemes

Sample NetFlow Enabling Schemes

Note Enabling NetFlow on a per interface basis is not supported on a Catalyst 4500 switch.

This example shows how to enable NetFlow globally:

Switch# configure terminal

Switch(config)# ip flow ingress

This example shows how to enable NetFlow with support for inferred fields:

Switch# configure terminal

Switch(config)# ip flow ingress infer-fields

Sample NetFlow Aggregation Configurations

This section provides the following aggregation cache configuration examples:

•Autonomous System Configuration

•Destination Prefix Configuration

•Prefix Configuration

•Protocol Port Configuration

•Source Prefix Configuration

Autonomous System Configuration

This example shows how to configure an autonomous system aggregation cache with an inactive timeout of 200 seconds, a cache active timeout of 45 minutes, an export destination IP address of 10.42.42.1, and a destination port of 9992:

Switch(config)# ip flow-aggregation cache as

Switch(config-flow-cache)# cache timeout inactive 200

Switch(config-flow-cache)# cache timeout active 45

Switch(config-flow-cache)# export destination 10.42.42.1 9992

Switch(config-flow-cache)# enabled

Destination Prefix Configuration

This example shows how to configure a destination prefix aggregation cache with an inactive timeout of 200 seconds, a cache active timeout of 45 minutes, an export destination IP address of 10.42.42.1, and a destination port of 9992:

Switch(config)# ip flow-aggregation cache destination-prefix

Switch(config-flow-cache)# cache timeout inactive 200

Switch(config-flow-cache)# cache timeout active 45

Switch(config-flow-cache)# export destination 10.42.42.1 9992

Switch(config-flow-cache)# enabled

Prefix Configuration

This example shows how to configure a prefix aggregation cache with an inactive timeout of 200 seconds, a cache active timeout of 45 minutes, an export destination IP address of 10.42.42.1, and a destination port of 9992:

Switch(config)# ip flow-aggregation cache prefix

Switch(config-flow-cache)# cache timeout inactive 200

Switch(config-flow-cache)# cache timeout active 45

Switch(config-flow-cache)# export destination 10.42.42.1 9992

Switch(config-flow-cache)# enabled

Protocol Port Configuration

This example shows how to configure a protocol port aggregation cache with an inactive timeout of 200 seconds, a cache active timeout of 45 minutes, an export destination IP address of 10.42.42.1, and a destination port of 9992:

Switch(config)# ip flow-aggregation cache protocol-port

Switch(config-flow-cache)# cache timeout inactive 200

Switch(config-flow-cache)# cache timeout active 45

Switch(config-flow-cache)# export destination 10.42.42.1 9992

Switch(config-flow-cache)# enabled

Source Prefix Configuration

This example shows how to configure a source prefix aggregation cache with an inactive timeout of 200 seconds, a cache active timeout of 45 minutes, an export destination IP address of 10.42.42.1, and a destination port of 9992:

Switch(config)# ip flow-aggregation cache source-prefix

Switch(config-flow-cache)# cache timeout inactive 200

Switch(config-flow-cache)# cache timeout active 45

Switch(config-flow-cache)# export destination 10.42.42.1 9992

Switch(config-flow-cache)# enabled

Sample NetFlow Minimum Prefix Mask Router-Based Aggregation Schemes

This section provides examples for the NetFlow minimum prefix mask aggregation cache configuration:

•Prefix Aggregation Scheme

•Destination-Prefix Aggregation Scheme

•Source-Prefix Aggregation Scheme

Prefix Aggregation Scheme

This is an example of a prefix aggregation cache configuration:

!

ip flow-aggregation cache prefix

mask source minimum 24

mask destination minimum 28

In this example, assume the following configuration:

ip route 118.42.20.160 255.255.255.224 110.42.13.2

ip route 122.16.93.160 255.255.255.224 111.22.21.2

Both routes have a 27-bit subnet mask in the routing table on the switch.

Flows travelling from the 118.42.20.160 subnet to the 122.16.93.160 subnet whose source IP addresses match with a mask of 27 bits and whose destination IP addresses match with a mask of 28 bits are aggregated together in the cache statistics.

Destination-Prefix Aggregation Scheme

This is an example of a destination-prefix aggregation cache configuration:

!

ip flow-aggregation cache destination-prefix

mask destination minimum 32

!

Source-Prefix Aggregation Scheme

This is an example of a source-prefix aggregation cache configuration:

ip flow-aggregation cache source-prefix

mask source minimum 30

!