Explains how hardware-offloaded loopback probes measure IPv6 endpoint latency, loss, liveness, and ECMP paths without remote monitoring support or PTP clock synchronization.
A loopback-based performance monitoring mechanism for IPv6 endpoints is an SRv6 network monitoring capability that
-
measures round-trip latency, packet loss, and liveness for IPv6 endpoints
-
uses a loopback mechanism to remove the need for performance monitoring or STAMP support on the destination endpoint, and
-
operates without Precision Time Protocol (PTP) synchronization between the sender and destination endpoints.
|
Feature Name |
Release Information |
Feature Description |
|---|---|---|
|
Loopback-based performance monitoring for IPv6 endpoints |
Release 26.3.1 |
Introduced in this release on: Fixed Systems (8200 [ASIC: Q200, P100], 8700 [ASIC: P100, K100], 8010 [ASIC: A100]); Centralized Systems (8600 [ASIC: Q200]); Modular Systems (8800 [LC ASIC: Q200, P100]) You can now simplify IPv6 endpoint performance monitoring by eliminating the need to configure remote devices, enable IPM or STAMP support on those endpoints, or synchronize clocks with PTP. IPM loopback mode uses hardware-offloaded SRv6 probes to measure latency, packet loss, and liveness, and uses flow-label spraying to validate available Equal-Cost Multipath (ECMP) paths. The feature provides high-precision performance metrics through hardware-based timestamping and exports real-time results through telemetry and syslog alerts. |
Network visibility gaps in IPv6 endpoint monitoring
Traditional one-way Integrated Performance Measurement (IPM) deployments require the destination endpoint to participate in measurement processing and often require PTP synchronization between the source and destination. These requirements can limit performance monitoring for legacy devices, third-party endpoints, or endpoints that cannot run active measurement software.
In an ECMP environment, standard connectivity checks may validate only one network path, which can hide degradation on other ECMP paths. Loopback mode helps address this gap by using sender-hosted probes that can be distributed across ECMP paths without requiring remote endpoint configuration or clock synchronization.
The sender node hosts the IPM loopback session and offloads probe generation, timestamping, and metric calculation to the NPU. Only liveness is calculated in real time. Latency and packet loss are calculated after each alternate-marking interval.
Benefits of loopback-based performance monitoring for IPv6 endpoints
Loopback-based performance monitoring offers these benefits:
-
Simplified deployment: Monitors legacy or third-party IPv6 endpoints without requiring IPM, STAMP, endpoint configuration, or PTP synchronization on the destination.
-
Comprehensive path coverage: Uses flow-label spraying to distribute probes across ECMP paths and detect path-specific degradation.
-
High-scale performance: Offloads probe generation, timestamping, and metric calculation to the NPU to preserve system processor resources.
-
High-precision metrics: Uses hardware timestamping at NPU ingress and egress points to calculate latency and packet-loss metrics.
-
Seamless observability: Exports real-time performance data through telemetry and syslog alerts for existing monitoring systems.
IPM loopback and IPM one-way comparison
|
Feature |
IPM loopback |
IPM one-way |
|---|---|---|
|
Peer requirement |
The destination endpoint does not need to support IPM or STAMP. |
The destination endpoint must be an active IPM-capable receiver. |
|
PTP synchronization |
PTP synchronization is not required between the sender and destination endpoint. |
PTP synchronization is required between the sender and destination endpoint. |
|
Processing location |
Latency, loss, and liveness calculations occur on the sender node. The receiver does not perform measurement processing. |
The sender and receiver participate in timestamping and processing. |
|
Scalability |
Sessions can be hosted for endpoints that do not support active receiver capability, because the destination endpoint only needs to return the probes through the configured reverse path. |
Scale is limited by the requirement for active receiver capability at the destination endpoint. |
|
Measurement scope |
Measures aggregate round-trip performance across the forward and reverse paths. |
Measured in one direction only. |