SRv6 Flow Label entropy via source address
SRv6 Flow Label entropy via source address is a core segment routing feature that
-
enables consistent end-to-end load balancing in networks when intermediate nodes are unable to utilize the standard IPv6 Flow Label for hash calculations
-
serves as a critical workaround for service providers and cloud-based deployments, and
-
provides flexible configuration options allowing for platform-specific optimization.
|
Feature Name |
Release |
Description |
|---|---|---|
|
SRv6 Flow Label entropy via source address |
Release 26.2.1 |
Introduced in this release on: Fixed Systems (8200 [ASIC: P100], 8700 [ASIC: P100, K100], 8010 [ASIC: A100]); Centralized Systems (8400 [ASIC: K100]); Modular Systems (8800 [LC ASIC: P100]) You can now improve SRv6 load balancing by encoding entropy from the IPv6 Flow Label into the IPv6 source address during SRv6 encapsulation. The ingress node generates a flow-aware source address. This allows downstream devices to use standard IP-based hashing for consistent ECMP load balancing. With this feature, devices use the modified source address for hashing so effective end-to-end load balancing occurs, even when the flow label is not considered. Previously, SRv6 relied on Flow Label–based entropy for load balancing. Some backbone nodes, especially in cloud environments ignore the Flow Label. This causes traffic polarization and inefficient bandwidth utilization. The feature introduces these changes: CLI:
|
SRv6 entropy encoding in IPv6 source address
In an SRv6 network, the Flow Label in the IPv6 header provides built-in entropy and enables efficient per-flow load balancing. the core nodes use this entropy to perform hashing without accessing all flows arriving at the PE.
Some IPv6 backbone nodes, especially in cloud environments do not use the Flow Label for load balancing, which disrupts end-to-end traffic distribution. To address this, the encapsulating node encodes entropy into the IPv6 source address during SRv6 H.Encaps. It generates a flow-aware source address by combining the Flow Label with the original source address, enabling all network devices to use standard IP-based hashing and ensures consistent load balancing across the network.
Benefits of Flow Label entropy via source address
SRv6 Flow Label entropy using source address includes the listed benefits.
-
Ensures effective ECMP distribution across all network segments, including cloud underlays that ignore the Flow Label.
-
Prevents traffic polarization and maximizes use of available bandwidth.
-
Enables reliable SRv6 deployments over public cloud infrastructures.
-
Eliminates reliance on intermediate nodes supporting Flow Label-based hashing.
-
Reduces congestion and improves overall application and service reliability.
Usage guidelines of Flow Label entropy via source address
These guidelines and limitations apply when you configure Flow Label entropy via source address:
-
This feature supports both SRv6 full-length SID and micro-SIDs.
-
The Flow Label uses a fixed size of 16 bits.
-
The feature does not support offset or size selection for flexible entropy encoding.
-
Use the
right-justifiedmode for Flow Label positioning. -
Only one source address (explicit or locator-derived) can be configured at a time.
-
If you configure an invalid source address or if the locator is down, the feature programs a null source address. It does not revert to the default source address.
How SRv6 Flow Label entropy via source address works
Summary
Embedding SRv6 Flow Label entropy into the IPv6 source address allows core network nodes to leverage source address diversity for effective load balancing, even when Flow Label–based entropy is not available.
Key components of the process include:
-
Ingress PE device: Receives the original packet and prepares it for SRv6 encapsulation.
-
Hardware hash function: Calculates the entropy value (Flow Label) based on packet headers.
-
Core node: Performs load-balancing decisions using the entropy in the modified source address.
-
Egress PE: Removes the SRv6 header and delivers the original payload.
Workflow
These stages describe how SRv6 Flow Label entropy via source address works:
- The ingress PE receives the data packet.
- The hardware computes a hash from the packet headers and generates the Flow Label.
- The PE applies the SRv6 H.Encaps operation and adds the SRv6 header.
- The ingress PE uses the configured add-entropy parameters to encode Flow Label bits into the IPv6 source address at the required offset.
- The ingress PE forwards the packet into the core network. The core nodes use the modified source address to perform load-balancing hashes.
- The egress PE terminates the SRv6 header and forwards the original payload to the destination.
Configure Flow Label entropy via source address
The configuration enables the router to use a specified locator as the source address. It injects Flow Label bits into this address to enhance load balancing performance in the network.
Follow these steps to configure SRv6 Flow Label entropy via source address.
Before you begin
Ensure that the SRv6 locator is configured and operational.
Procedure
|
Step 1 |
Run the segment-routing srv6 encapsulation command to configure the encapsulation parameters for SRv6 traffic. Example:
|
|
Step 2 |
Configure the router to encode Flow Label–based entropy into the IPv6 source address using a locator.
|
|
Step 3 |
Enter these show commands to verify that Flow Label–based entropy is encoded in the IPv6 source address. Example:
|
Feedback