Update the dataplane hashing test README to cover more traffic profile and test scenario - #5954
Update the dataplane hashing test README to cover more traffic profile and test scenario#5954singhh07 wants to merge 4 commits into
Conversation
…ap, and IPnIP Decap traffic flows, along with WCMP and asymmetric capacity
## Description
This PR updates the test plan and specification in `feature/gribi/otg_tests/hashing_test/README.md` to comprehensively cover dataplane hashing test cases, including tunnel encapsulation/decapsulation handling, weighted cost multi-pathing (WCMP), and asymmetric Link Aggregation Group (LAG) member capacity.
## Key Updates
1. **Traffic Profiles**:
- **Plain IPv4/IPv6**: Evaluates native 5-tuple L3/L4 entropy.
- **IPnIP Encapsulation (Encap)**: Uses static outer IP endpoints (0 outer entropy) with randomized 5-tuple inner payload to verify inner-header hash computation and prevent tunnel polarization.
- **IPnIP Decapsulation (Decap)**: Matches gRIBI decapsulation routes to verify post-decap inner-header hashing.
2. **Tolerance and Evaluation Criteria**:
- Standardized the acceptable distribution threshold to **$\pm 2\%$ relative tolerance** ($\text{Expected Ratio} \times (1 \pm 0.02)$).
3. **Test Scenarios**:
- **Scenario 1 (8-Wide ECMP & WCMP)**:
- *Sub-case 1.1*: 8-Wide Uniform ECMP ($\sim 12.5\%$ per next-hop).
- *Sub-case 1.2*: Equal Paths, Unequal Weights ($1:2$ ratio $\rightarrow \sim 8.3\%$ on soft loops vs. $\sim 16.6\%$ on physical loops).
- **Scenario 2 (Intra-LAG Member Distribution)**:
- Uniform distribution across a 7-member LAG bundle ($\sim 14.28\%$ per link).
- **Scenario 3 (Asymmetric Paths & Weighted Load Balancing — 3-Wide LAGs)**:
- *Sub-case 3.1 (Capacity-Based Weights — 3:2:2)*: Matches physical member counts ($\sim 42.86\%, \sim 28.57\%, \sim 28.57\%$).
- *Sub-case 3.2 (Overriding Capacity / Equal Weights — 1:1:1)*: Verifies software weights strictly override physical member capacity ($\sim 33.33\%$ split per LAG).
4. **OpenConfig Coverage**:
- Added OpenConfig telemetry paths for loopback-mode, aggregate LAG attributes, ACL bindings, and gRIBI / gNMI RPCs.
Summary of ChangesHello, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request significantly enhances the dataplane hashing test specification for gRIBI/OTG environments. By introducing diverse traffic profiles and complex test scenarios, the update ensures robust validation of hashing uniformity, weight enforcement, and anti-polarization mechanisms across various network instances and LAG configurations. Highlights
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize the Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counterproductive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here. Footnotes
|
Pull Request Functional Test Report for #5954 / 56ad966Virtual Devices
Hardware Devices
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request comprehensively updates the dataplane hashing test documentation (README.md), detailing the topology of physical and software loopbacks, traffic profiles (Plain IP, IPnIP Encap, and IPnIP Decap), and three new test scenarios covering multi-stage fan-out, intra-LAG distribution, and asymmetric paths. Feedback on the changes highlights a mathematical error in the expected traffic distribution percentages for Stage 3 in Sub-case 1.2, where the ratio of physical to software loopbacks differs from Stage 2.
…e and test scenario
## Summary
This PR updates the test plan specification in `feature/gribi/otg_tests/hashing_test/README.md` to define Dataplane Hashing verification (ECMP, WCMP, and Intra-LAG load balancing) using a combination of physical loopback pairs and software terminal loopback interfaces across multiple Network Instances (`DEFAULT`, `TRANSIT`, `SELF_SITE`, `EGRESS`) with different traffic profiles.
---
## 1. Traffic Profiles
- **Plain IPv4/IPv6**: Evaluates native 5-tuple L3/L4 entropy across destination IP, source IP, and UDP ports.
- **IPnIP Encapsulation (Encap)**: Uses static outer IP endpoints (0 outer entropy) with randomized 5-tuple inner payload to verify inner-header hash computation and prevent tunnel polarization.
---
## 2. Tolerance & Evaluation Criteria
Standardized the acceptable distribution threshold across all next-hops and member links to a **$\pm 2\%$ relative tolerance**:
$$\text{Acceptable Ratio Range} = \text{Expected Ratio} \times (1 \pm 0.02)$$
---
## 3. Test Scenarios
### **Scenario 1: Multi-Stage Max Fan-out (8-Wide ECMP & WCMP)**
- **Sub-case 1.1: 8-Wide Uniform ECMP**:
- Uniform distribution across all 8 next-hops (**~12.50%** each, acceptable range: **12.25% – 12.75%**) across `TRANSIT` and `SELF_SITE` VRFs.
- **Sub-case 1.2: Equal Paths, Unequal Weights (1:2 WCMP Ratio)**:
- **Stage 2 (`TRANSIT`)** (4 Physical Loops @ weight 2 + 4 Soft Loops @ weight 1 $\rightarrow$ Total weight 12):
- **Physical Loops (4 ports)**: **~16.67%** each (acceptable range: **16.33% – 17.00%**)
- **Soft Loops (4 ports)**: **~8.33%** each (acceptable range: **8.16% – 8.50%**)
- **Stage 3 (`SELF_SITE`)** (3 Physical Loops @ weight 2 + 5 Soft Loops @ weight 1 $\rightarrow$ Total weight 11):
- **Physical Loops (3 ports)**: **~18.18%** each (acceptable range: **17.82% – 18.54%**)
2 participants
Update the dataplane hashing test README to cover more traffic profile and test scenario
Summary
This PR updates the test plan specification in
feature/gribi/otg_tests/hashing_test/README.mdto define Dataplane Hashing verification (ECMP, WCMP, and Intra-LAG load balancing) using a combination of physical loopback pairs and software terminal loopback interfaces across multiple Network Instances (DEFAULT,TRANSIT,SELF_SITE,EGRESS) with different traffic profiles.1. Traffic Profiles
2. Tolerance & Evaluation Criteria
Standardized the acceptable distribution threshold across all next-hops and member links to a$\pm 2%$ relative tolerance:
$$\text{Acceptable Ratio Range} = \text{Expected Ratio} \times (1 \pm 0.02)$$
3. Test Scenarios
Scenario 1: Multi-Stage Max Fan-out (8-Wide ECMP & WCMP)
TRANSITandSELF_SITEVRFs.TRANSIT) (4 Physical Loops @ weight 2 + 4 Soft Loops @ weight 1SELF_SITE) (3 Physical Loops @ weight 2 + 5 Soft Loops @ weight 1This PR updates the test plan and specification in
feature/gribi/otg_tests/hashing_test/README.mdto comprehensively cover dataplane hashing test cases, including tunnel encapsulation/decapsulation handling, weighted cost multi-pathing (WCMP), and asymmetric Link Aggregation Group (LAG) member capacity. ## Key Updates