EWMA score adjusted by success rate when backend services fail fast

[jizhuozhi]

What this PR does / why we need it:

PeakEWMA (or EWMA) is a latency-based load balancing algorithm. Latency-based load balancing algorithms are all based on the premise that the cost of processing requests is the same. But as we all know, user input is untrustworthy, and all backend services will verify user (or other service) input, and once the verification fails, the request will not be processed and the result will be returned quickly, which will lead to The load balancing algorithm gets a wrong view: this request is processed very quickly.

$$ \begin{align} Avg &= (\sum_{i=1}^{SuccessCount}SuccessDuration_i + \sum_{i=1}^{FailedCount}FailedDuration_i) / TotalCount \\ &\approx \sum_{i=1}^{SuccessCount}SuccessDuration_i / TotalCount \\ &\ll \sum_{i=1}^{SuccessCount}SuccessDuration_i / SuccessCount \end{align} $$

The same problem is also described in "The Site Reliability Workbook: Practical Ways to Implement SRE". When the resource-based load balancing algorithm encounters fail-fast, traffic will skew to the abnormal server.

We need a mathematical model to fix this wrong view. The easiest ways to do this is to fix all response times to the same value as the SuccessDuration

$$ \begin{align} FixedAvg &= Avg / \alpha \\ SuccessCount &= \alpha TotalCount \\ \alpha &= SuccessCount / TotalCount \end{align} $$

Why use EWMA to count the failure rate instead of the arithmetic mean:

The number of long-running service samples will continue to increase. The longer the running time, the harder it is to detect short-term failures. At the same time, because the traffic distribution will be reduced after the failure is detected, it will take a long time to recover enough samples. And through EWMA, it can automatically decay to a small enough value over time

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• CVE Report (Scanner found CVE and adding report)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation only

Which issue/s this PR fixes

How Has This Been Tested?

All effects are limited to ewma load balancer, so I only added test cases for EWMA. I mocked the failed response and tested for changes in the EWMA value.

Checklist:

• My change requires a change to the documentation.
• I have updated the documentation accordingly.
• I've read the CONTRIBUTION guide
• I have added unit and/or e2e tests to cover my changes.
• All new and existing tests passed.
• Added Release Notes.

Does my pull request need a release note?

Any user-visible or operator-visible change qualifies for a release note. This could be a:

• CLI change
• API change
• UI change
• configuration schema change
• behavioral change
• change in non-functional attributes such as efficiency or availability, availability of a new platform
• a warning about a deprecation
• fix of a previous Known Issue
• fix of a vulnerability (CVE)

No release notes are required for changes to the following:

• Tests
• Build infrastructure
• Fixes for unreleased bugs

For more tips on writing good release notes, check out the Release Notes Handbook

EWMA score adjusted by success rate when backend services fail fast (@jizhuozhi)

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: jizhuozhi / name: 纪卓志 (7b64e47)

[k8s-ci-robot]

This issue is currently awaiting triage.

If Ingress contributors determines this is a relevant issue, they will accept it by applying the triage/accepted label and provide further guidance.

The triage/accepted label can be added by org members by writing /triage accepted in a comment.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[k8s-ci-robot]

Welcome @jizhuozhi!

It looks like this is your first PR to kubernetes/ingress-nginx 🎉. Please refer to our pull request process documentation to help your PR have a smooth ride to approval.

You will be prompted by a bot to use commands during the review process. Do not be afraid to follow the prompts! It is okay to experiment. Here is the bot commands documentation.

You can also check if kubernetes/ingress-nginx has its own contribution guidelines.

You may want to refer to our testing guide if you run into trouble with your tests not passing.

If you are having difficulty getting your pull request seen, please follow the recommended escalation practices. Also, for tips and tricks in the contribution process you may want to read the Kubernetes contributor cheat sheet. We want to make sure your contribution gets all the attention it needs!

Thank you, and welcome to Kubernetes. 😃

[k8s-ci-robot]

Hi @jizhuozhi. Thanks for your PR.

I'm waiting for a kubernetes member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[strongjz]

/ok-to-test

[strongjz]

@jizhuozhi the lua tests failed, you can check the prow job logs for the reasons.

/hold

[jizhuozhi]

@jizhuozhi the lua tests failed, you can check the prow job logs for the reasons.

Thanks, I will check the failed jobs and fix them

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: jizhuozhi
Once this PR has been reviewed and has the lgtm label, please assign puerco for approval. For more information see the Kubernetes Code Review Process.

The full list of commands accepted by this bot can be found here.

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[jizhuozhi]

Hello, @strongjz , failed tests are fixed, PLAT 😺

[jizhuozhi]

Hello, @strongjz @tao12345666333 , PTAL, thanks

[tao12345666333]

/assign