Specify load-balancing behavior for each service that is being proxied

[hbagdi]

What would you like to be added:

For any service that is exposed via service-apis, a service owner should be able to specify the load-balancing behavior that should be used. This can include a common set of algorithms like round-robin, weighted, hash based routing, cookie/session based routing.

Why is this needed:

Most services in Kubernetes are designed to scale horizontally. A lot of legacy services that are ported to run on Kubernetes have the requirement of stick sessions. Most proxies and load balancers support load-balancing as well and this is a widely adopted feature in most of current ingress implementations.

[hbagdi]

There is certainly an overlap with kube-proxy and the corev1.Service resource here.

[bowei]

Yes, this API would give us an opportunity to add better load balancing knobs without adding more to the core.Service bucket

[hbagdi]

@robscott Seems like there is already an issue open for load-balancing.

[hbagdi]

/kind user-story

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[jpeach]

/remove-lifecycle stale
/lifecycle frozen

[hbagdi]

In search for a solution, we were unsuccessful in coming up with a portable way to support this feature.
As a compromise, we came up with an easy and portable way to support such implementation-specific features in a consistent manner. The solution is documented in #713.

If there are more portable ways to tackle this, we would love to hear about them.

For now, I'll close this issue.
/close

[k8s-ci-robot]

@hbagdi: Closing this issue.

In response to this:

In search for a solution, we were unsuccessful in coming up with a portable way to support this feature.
As a compromise, we came up with an easy and portable way to support such implementation-specific features in a consistent manner. The solution is documented in #713.

If there are more portable ways to tackle this, we would love to hear about them.

For now, I'll close this issue.
/close

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.