Configuring IP FPM Hop-by-Hop Performance Statistics Collection
IP Flow Performance Monitor (FPM) hop-by-hop performance statistics collection helps locate faulty nodes when packet loss or delay occurs.
Usage Scenario
- When a target flow is unidirectional, you can directly implement hop-by-hop performance statistics collection for the flow.
- When a target flow is bidirectional, two situations are available:
- If the target flow is symmetrical, you can implement hop-by-hop performance statistics collection for the forward or backward flow, and the measurement is the same either way.
- If the target flow is asymmetrical, you must implement hop-by-hop performance statistics collection for both the forward and backward flows to obtain their respective measurements.
These measurements serve as a reliable reference for network operation and maintenance and fault diagnosis, improving network reliability and user experience.
Pre-configuration Tasks
Before configuring IP FPM hop-by-hop performance statistics collection, complete the following tasks:
Configure a dynamic routing protocol or static routes so that devices are reachable at the network layer.
Configure the network time protocol (NTP) or 1588v2 so that all device clocks can be synchronized.
NOTE:
Compared with NTP, 1588v2 implements inter-device clock synchronization with higher precision. To obtain accurate performance statistics, it is recommended that 1588v2 be deployed to implement clock synchronization between all devices with clocks when you use IP FPM for performance measurement, especially for one-way delay measurement.
- Configuring an MCP
A Measurement Control Point (MCP) collects statistics reported by Data Collecting Points (DCPs), summarizes and calculates the statistics, and reports measurement results to user terminals or the network management system (NMS). - Configuring a DCP
A Data Collecting Point (DCP) manages and controls Target Logical Ports (TLPs), collects statistics generated by TLPs, and reports the statistics to a Measurement Control Point (MCP). - Checking the Configurations
After configuring the MCP, DCPs, and TLPs, check the IP FPM hop-by-hop performance statistics collection function.
Configuring an MCP
A Measurement Control Point (MCP) collects statistics reported by Data Collecting Points (DCPs), summarizes and calculates the statistics, and reports measurement results to user terminals or the network management system (NMS).
Context
Device A functions as an MCP to collect statistics reported by DCP1, DCP2, and DCP3, summarize and calculate the statistics, and report measurement results to user terminals or the NMS.
Perform the following steps on Device A:
Procedure
- Run system-view
The system view is displayed.
- Run nqa ipfpm mcp
MCP is enabled globally, and the IPFPM-MCP view is displayed.
- Run mcp id mcp-id
An MCP ID is configured.
Using the Router ID of a device that is configured as an MCP as its MCP ID is recommended.
The MCP ID must be an IP address reachable to DCPs. The MCP ID configured on an MCP must be the same as that specified in the mcp mcp-id [ port port-number ] command run in the IP FPM instance view of all DCPs associated with this MCP. If an MCP ID is changed on an MCP, it must be changed for all DCPs associated with this MCP in an IP FPM instance. Otherwise, the MCP cannot process the statistics reported by the DCPs.
- (Optional) Run protocol udp port port-number
A UDP port number is specified for the MCP to communicate with DCPs.
The UDP port number configured on an MCP must be the same as that specified in the mcp mcp-id [ port port-number ] command run in the IP FPM instance view of all DCPs associated with this MCP. If a UDP port number is changed on an MCP, it must be changed for all DCPs associated with this MCP in an IP FPM instance. Otherwise, the MCP cannot process the statistics reported by the DCPs.
- (Optional) Run authentication-mode hmac-sha256 key-id key-id [ cipher ] password
The authentication mode and password are configured on the MCP.
The authentication mode and password configured on an MCP must be the same as those configured in the authentication-mode hmac-sha256 key-id key-id [ cipher ] password command run on all DCPs associated with this MCP. Otherwise, the MCP cannot process the statistics reported by the DCPs.
- Run instance instance-id
An IP FPM instance is created, and the instance view is displayed.
instance-id must be unique on an MCP and all its associated DCPs. The MCP and all its associated DCPs must have the same IP FPM instance configured. Otherwise, statistics collection does not take effect.
- (Optional) Run description text
The description is configured for the IP FPM instance.
The description of an IP FPM instance can contain the functions of the instance, facilitating applications.
- Run dcp dcp-id
A DCP ID is specified in the IP FPM instance.
The DCP ID configured in an IP FPM instance must be the same as that specified in the dcp id dcp-id command run on a DCP. Otherwise, the MCP associated with this DCP cannot process the statistics reported by the DCP.
- Run the following commands to configure Atomic Closed Hops
(ACHs).
An ACH identifies a range between two neighboring measurement points. The network shown in Figure 2-7 is classified into three ACHs: ACH1 {TLP100, TLP200}, ACH2 {TLP200, TLP300}, and ACH3 {TLP200, TLP310}. In ACH1, TLP100 is the in-point, and TLP200 is the out-point. In ACH2, TLP200 is the in-point, and TLP300 is the out-point. In ACH3, TLP200 is the in-point, and TLP310 is the out-point.
- Run the ach ach-id command to create an ACH and enter the ACH view.
- Run the flow { forward | backward | bidirectional } command to specify the direction in which hop-by-hop delay measurement is implemented for the target flow.
Run the in-group dcp dcp-id tlp tlp-id command to configure the TLP in-group.
Run the out-group dcp dcp-id tlp tlp-id command to configure the TLP out-group.
Follow-up Procedure
When DCP configurations are being changed, the MCP may receive incorrect statistics from the DCP. To prevent this, run the measure disable command to disable IP FPM performance statistics collection of a specified instance on the MCP. After the DCP configuration change is complete, run the undo measure disable or measure enable command to enable IP FPM performance statistics collection for the specified instance on the MCP. This ensures accurate measurement.
Configuring a DCP
A Data Collecting Point (DCP) manages and controls Target Logical Ports (TLPs), collects statistics generated by TLPs, and reports the statistics to a Measurement Control Point (MCP).
Context
Device A, Device B, and Device C function as DCPs. Device A manages and controls TLP100, Device B manages and controls TLP200, and Device C manages and control TLP300 and TLP310. Device A, Device B, and Device C collect statistics generated by these TLPs and report the statistics to the MCP.
Perform the following steps on Device A, Device B, and Device C:
Procedure
- Run system-view
The system view is displayed.
- Run nqa ipfpm dcp
DCP is enabled globally, and the IPFPM-DCP view is displayed.
- Run dcp id dcp-id
A DCP ID is configured.
Using the Router ID of a device that is configured as a DCP as its DCP ID is recommended.
The DCP ID configured on a DCP must be the same as that specified in the dcp dcp-id command run in the IP FPM instance view of the MCP associated with this DCP. Otherwise, the MCP cannot process the statistics reported by the DCP.
- (Optional) Run authentication-mode hmac-sha256 key-id key-id [ cipher ] password
The authentication mode and password are configured on the DCP.
The authentication mode and password configured on a DCP must be the same as those configured in the authentication-mode hmac-sha256 key-id key-id [ cipher ] password command run on the MCP associated with the DCP. Otherwise, the MCP cannot process the statistics reported by the DCP.
- (Optional) Run color-flag loss-measure { tos-bit tos-bit | flags-bit0 } delay-measure { tos-bit tos-bit | flags-bit0 }
IP FPM measurement flags are configured.
The loss and delay measurement flags cannot use the same bit, and the bits used for loss and delay measurement must not have been used in other measurement tasks.
- Run mcp mcp-id [ port port-number ] [ vpn-instance vpn-instance-name | net-manager-vpn ]
An MCP ID is specified for the DCP, and the UDP port number is configured for the DCP to communicate with the MCP.
The UDP port number configured on the DCP must be the same as that configured in the protocol udp port port-number command run on the MCP associated with this DCP. Otherwise, the DCP cannot report the statistics to the MCP.
The VPN instance has been created on the DCP before you configure vpn-instance vpn-instance-name or net-manager-vpn to allow the DCP to report the statistics to the MCP through the specified VPN or management VPN.
- (Optional) Run period source ntp
The DCP is configured to select NTP as the clock source when calculating an IP FPM statistical period ID.
In P2MP (MP being two points) delay measurement scenarios, if the ingress of the service traffic uses NTP as the clock source, but the egresses use a different clock source, for example, NTP or 1588v2, you must configure the egresses to select NTP as the clock source when calculating an IP FPM statistical period ID to ensure consistent clock sources on the ingress and egresses.
- Run instance instance-id
An IP FPM instance is created, and the instance view is displayed.
instance-id must be unique on an MCP and all its associated DCPs. The MCP and all its associated DCPs must have the same IP FPM instance configured. Otherwise, statistics collection does not take effect.
- (Optional) Run description text
The description is configured for the IP FPM instance.
The description of an IP FPM instance can contain the functions of the instance, facilitating applications.
- (Optional) Run interval interval
The statistical period is configured for the IP FPM instance.
- Perform either of the following operations to configure
the target flow characteristics in the IP FPM instance.
Configure the forward or backward target flow characteristics.
When protocol is specified as TCP or UDP, run:
flow { forward | backward } { protocol { tcp | udp } { source-port src-port-number1 [ to src-port-number2 ] | destination-port dest-port-number1 [ to dest-port-number2 ] } * | dscp dscp-value | source src-ip-address [ src-mask-length ] | destination dest-ip-address [ dest-mask-length ] } *
When protocol is specified as any protocol other than TCP or UDP, run:
flow { forward | backward } { protocol protocol-number | dscp dscp-value | source src-ip-address [ src-mask-length ] | destination dest-ip-address [ dest-mask-length ] } *
Configure the characteristics for the bidirectional target flow.When protocol is specified as TCP or UDP, run:
flow bidirectional { protocol { tcp | udp } { source-port src-port-number1 [ to src-port-number2 ] | destination-port dest-port-number1 [ to dest-port-number2 ] } * | dscp dscp-value | source src-ip-address [ src-mask-length ] | destination dest-ip-address [ dest-mask-length ] } *
When protocol is specified as any protocol other than TCP or UDP, run:
flow bidirectional { protocol protocol-number | dscp dscp-value | source src-ip-address [ src-mask-length ] | destination dest-ip-address [ dest-mask-length ] } *
NOTE:
If the target flow in an IP FPM instance is unidirectional, only forward can be specified.
- If the target flow in an IP FPM instance is bidirectional, two situations are available:
- If the bidirectional target flow is asymmetrical, you must configure forward and backward in two command instances to configure the characteristics for the forward and backward flows, respectively.
- If the bidirectional target flow is symmetrical, you can specify bidirectional to configure the bidirectional target flow characteristics. By default, the characteristics specified are used for the forward flow, and the reverse of those are used for the backward flow. Specifically, the source and destination IP addresses and port numbers specified for the forward flow are used respectively as the destination and source IP addresses and port numbers for the backward flow. If the target flow is symmetrical bidirectional, set src-ip-address to specify a source IP address and dest-ip-address to specify a destination IP address for the target flow.
- Run the following commands to configure TLPs.
Run the tlp { in-point | out-point } { ingress | egress } [ vpn-label vpn-label ] [ backward-vpn-label backward-vpn-label ] command to configure a TLP and specify it as an in-point or out-point.
On the network shown in Figure 2-8, TLP100 is the in-point, and TLP310 is the out-point.
Run the tlp tlp-id mid-point flow { forward | backward } { ingress | egress } [ vpn-label vpn-label [ lsp-label lsp-label [ lsp-label2 lsp-label2 ] ] [ flow-label ] [ control-word ] [ l2vpn [ tpid tpid ] ] ] command to configure a TLP and specify it as a mid-point. In a load-balancing scenario where different paths share the same interface or path segment, run the tlp tlp-id index index-id mid-point flow { forward | backward } { ingress | egress } vpn-label vpn-label [ lsp-label lsp-label [ lsp-label2 lsp-label2 ] [ flow-label ] [ control-word ] [ l2vpn [ tpid tpid ] ] ] command to configure the mid-point included in the IP FPM statistical instance and the role of the mid-point.
On the network shown in Figure 2-8, TLP200 and TLP300 are mid-points.
Run the tlp tlp-id mid-point flow bidirectional { ingress | egress } [ forward { vpn-label vpn-label [ lsp-label lsp-label [ lsp-label2 lsp-label2 ] ] [ flow-label ] [ control-word ] [ l2vpn [ tpid tpid ] ] }] [ backward { vpn-label vpn-label [ lsp-label lsp-label [ lsp-label2 lsp-label2 ] ] [ flow-label ] [ control-word ] [ l2vpn [ tpid tpid ] ] } ] command to configure a TLP and specify it as a mid-point for the bidirectional target flow. In a load-balancing scenario where different paths share the same interface or path segment, run the tlp tlp tlp-id index index-id mid-point flow bidirectional { ingress | egress } { forward vpn-label vpn-label [ lsp-label lsp-label [ lsp-label2 lsp-label2 ] ] [ flow-label ] [ control-word ] [ l2vpn [ tpid tpid ] ] | backward vpn-label vpn-label [ lsp-label lsp-label [ lsp-label2 lsp-label2 ] ] [ flow-label ] [ control-word ] [ l2vpn [ tpid tpid ] ] | forward vpn-label vpn-label [ lsp-label lsp-label [ lsp-label2 lsp-label2 ] ] [ flow-label ] [ control-word ] [ l2vpn [ tpid tpid ] ] backward vpn-label vpn-label [ lsp-label lsp-label [ lsp-label2 lsp-label2 ] ] [ flow-label ] [ control-word ] [ l2vpn [ tpid tpid ] ] } command to configure the mid-point of the bidirectional target flow included in the IP FPM statistical instance and the role of the mid-point.
Run the tlp tlp-id mid-point flow { forward | backward } { ingress | egress } [ vpn-label vpn-label [ lsp-label lsp-label lsp-label2 { lsp-label2 | any } ] ] [ flow-label ] [ control-word ] [ l2vpn [ tpid tpid ] ] } ] command or the tlp mid-point flow bidirectional { ingress | egress } [ forward { vpn-label vpn-label [ lsp-label lsp-label [ lsp-label2 { lsp-label2 | any } ] ] [ flow-label ] [ control-word ] [ l2vpn [ tpid tpid ] ] } ] [backward { vpn-label vpn-label [ lsp-label lsp-label [ lsp-label2 { lsp-label2 | any } ] ] [ flow-label ] [ control-word ] [ l2vpn [ tpid tpid ] ] } ] command to configure the TE label of the TLP in an IP FPM instance as a wildcard label.
NOTE:
- Label overlapping may occur when performance measurement is performed on packets carrying labels on the downstream interface. In this case, run the tlp nexthop command to specify a next hop address.
- If tlp-id is set to the same value in the preceding commands, the configurations of the command run later overwrite the configurations of the command run earlier.
- Run quit
Return to the IPFPM-DCP view.
- Run quit
Return to the system view.
- Bind the TLPs to interfaces.
Run the interface interface-type interface-name command to enter the interface view.
- Run either of the following commands:
- If the interface is a Layer 3 interface, run the ipfpm tlp tlp-id command.
- If the interface is a Layer 2 interface, run the ipfpm tlp tlp-id { ce-default-vlan | vlan-id vlan-id } command.
- Run the quit command to return to the system view.
- Configure IP FPM hop-by-hop performance statistics collection.
- Run the nqa ipfpm dcp command to enter the IPFPM-DCP view.
- Run the instance instance-id command to enter the IP FPM instance view.
- Run the loss-measure enable mid-point [ time-range time-range | continual ] command to enable hop-by-hop packet loss measurement.
Enable delay measurement.
If the target flow is unidirectional, run either of the following commands:To enable on-demand one-way delay measurement, run the delay-measure enable one-way tlp mid-point [ time-range time-range ] command.
To enable proactive one-way delay measurement, run the delay-measure enable one-way tlp mid-point continual command.
NOTE:
In NE20E implementation, hop-by-hop delay measurement can be performed after you enable this function on the TLP in-point (TLP100 in Figure 2-8) from which the target flow enters the network.
If the target flow is a bidirectional, run either of the following commands:To enable on-demand two-way delay measurement, run the delay-measure enable two-way tlp mid-point [ time-range time-range ] command.
To enable proactive two-way delay measurement, run the delay-measure enable two-way tlp mid-point continual command.
Checking the Configurations
After configuring the MCP, DCPs, and TLPs, check the IP FPM hop-by-hop performance statistics collection function.
Procedure
- Run the display ipfpm mcp command to check MCP configurations.
- Run the display ipfpm dcp command to check DCP configurations.
- Run the display ipfpm statistic-type { loss | oneway-delay | twoway-delay } instance instance-id ach ach-id command to check the hop-by-hop performance statistics for a specified ACH.
Example
Run the display ipfpm mcp command. The command output shows MCP configurations.
<HUAWEI> display ipfpm mcp
Specification Information:
Max Instance Number :64
Max DCP Number Per Instance :256
Max ACH Number Per Instance :16
Max TLP Number Per ACH :16
Configuration Information:
MCP ID : 1.1.1.1
Status : Active
Protocol Port : 65030
Total Instances : 10
Run the display ipfpmdcp command to view the DCP configuration in the IP FPM statistics system.
<HUAWEI> display ipfpm dcp
Specification Information(Main Board):
Max Instance Number :64
Max 10s Instance Number :64
Max 1s Instance Number :--
Max TLP Number :512
Max TLP Number Per Instance :8
Configuration Information:
DCP ID : 2.2.2.2
Loss-measure Flag : tos-bit6(default)
Delay-measure Flag : tos-bit7(default)
Authentication Mode : hmac-sha256
Test Instances MCP ID : 1.1.1.1
Test Instances MCP Port : 65030
Total Instances : 10
Run the displayipfpm statistic-type loss command to view the statistics about discarded packets on each node of an ACH in the IP FPM statistics system.
<HUAWEI> display ipfpm statistic-type loss instance 1 ach 1
Latest loss statistics of forward flow:
Unit: p - packet, b - byte
------------------------------------------------------------------------------------------
Period Loss(p) LossRatio(p) Loss(b) LossRatio(b)
------------------------------------------------------------------------------------------
136190088 10 10.000000% 1000 10.000000%
136190087 10 10.000000% 1000 10.000000%
136190086 10 10.000000% 1000 10.000000%
136190085 10 10.000000% 1000 10.000000%
136190084 10 10.000000% 1000 10.000000%
136190083 10 10.000000% 1000 10.000000%
136190082 10 10.000000% 1000 10.000000%
Latest loss statistics of backward flow:
Unit: p - packet, b - byte
------------------------------------------------------------------------------------------
Period Loss(p) LossRatio(p) Loss(b) LossRatio(b)
------------------------------------------------------------------------------------------
Run the display ipfpm statistic-type oneway-delay command to view the one-way delay statistics on each node of an ACH in the IP FPM statistics system.
<HUAWEI> display ipfpm statistic-type oneway-delay instance 1 ach 1
Latest one-way delay statistics of forward flow:
--------------------------------------------------
Period Delay(usec) Delay
Variation(usec)
--------------------------------------------------
136190120 100 0
136190119 100 0
136190118 100 0
136190117 100 0
136190116 100 0
136190115 100 0
136190114 100 0
Latest one-way delay statistics of backward flow:
--------------------------------------------------
Period Delay(usec) Delay
Variation(usec)
--------------------------------------------------
Run the display ipfpm statistic-type twoway-delay command to view the two-way delay statistics in a specific IP FPM statistics instance.
<HUAWEI> display ipfpm statistic-type twoway-delay instance 1
Latest two-way delay statistics:
--------------------------------------------------
TimeStamp Delay(usec) Delay
Variation(usec)
--------------------------------------------------
136118757 800 0
136118756 800 0
136118755 800 0
136118753 800 0
136118752 800 0
136118751 800 0
136118750 800 0
136118749 800 0
136118748 800 0
136118747 800 0
136118746 800 0
136118745 800 0
Latest one-way delay statistics of bidirectional flow:
--------------------------------------------------------------------------------
TimeStamp Forward ForwardDelay Backward BackwardDelay
Delay(usec) Variation(usec) Delay(usec) Variation(usec)
--------------------------------------------------------------------------------
136118757 400 0 400 0
136118756 400 0 400 0
136118755 400 0 400 0
136118753 400 0 400 0
136118752 400 0 400 0
136118751 400 0 400 0
136118750 400 0 400 0
136118749 400 0 400 0
136118748 400 0 400 0
136118747 400 0 400 0
136118746 400 0 400 0
136118745 400 0 400 0