Proposal to spin down WG Multi-tenancy

[Tasha Drew]

Hi all,

In the past 5 years, this group has made a lot of great progress towards meeting our goals. Initially, when this group spun up, we had the following charter (https://docs.google.com/document/d/1SkVdOPR4jozYDT8ro51hU3yrf1sHS8Gez73xM3PCsVo/edit#):

• Define the models of multi tenancy that Kubernetes will support. 

• Discuss, prioritize, and execute upon any remaining work that? needs to be done to support these models, both in terms of “control plane multi tenancy” and “node multi tenancy.” 

• Write guides and documentation for how to setup these models for a cluster.

• Discuss ways to make the models easier to deploy or default paths to do so.

• Create conformance tests that will prove that these models can be built and used in production environments.

Over time, we've achieved these goals, as we discussed at Kubecon Detroit (https://www.youtube.com/watch?v=JuCWTrsvSEw) last fall.

• Official docs: https://kubernetes.io/docs/concepts/security/multi-tenancy/
  
• Incubated and graduated the Hierarchical Namespace Controller: https://github.com/kubernetes-sigs/hierarchical-namespaces
• Incubated and graduated the virtual cluster project: https://github.com/kubernetes-sigs/cluster-api-provider-nested/tree/main/virtualcluster
• Created multi-tenant benchmarks to test your Kubernetes cluster for secure multi-tenant configuration: https://github.com/kubernetes-sigs/multi-tenancy/tree/master/benchmarks

After Kubecon Detroit, our leads (me, Adrian Ludwin, Ryan Bezdicek, Fei Guo, Jim Bugwadia) discussed that we had met our initial charter. We've now waited for six months and haven't seen new or emergent use cases that would require incubation or work from our group.

Working groups are intended to be time-limited, groups of people coming together to solve a cross-SIG, cross-functional challenge or question. I think at this point we can declare victory.

I'd like to thank everyone who's been part of our community for the past five years. This has been an awesome, friendly, focused, and productive group of people coming together from many different perspectives to answer a hard problem.

Based on lazy consensus, opening up the proposal to spin down the WG to the group. If we don't identify a reason _not_ to spin down, let's consider lazy consensus resolved in 2 weeks (April 2, 2023).

Thanks all,

Tasha

Co-Chair of the Multi-tenancy WG

[Adrian Ludwin]

Thanks Tasha. I second this - I think we've answered all the questions and built some great infrastructure to improve multitenancy in K8s. Thanks to Tasha for all your leadership!

Cheers, A

--
You received this message because you are subscribed to the Google Groups "kubernetes-wg-multitenancy" group.
To unsubscribe from this group and stop receiving emails from it, send an email to kubernetes-wg-multi...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/kubernetes-wg-multitenancy/ff6293b7-90fd-46a7-a3a4-5fde17e57eban%40googlegroups.com.

[David Oppenheimer]

Thanks Tasha, Adrian, and all the others who have led the MTWG (including notably Jessie Frazelle, who co-founded the group with me).

I think Kubernetes is in a very good position these days for "soft" multi-tenancy. I think there are still challenges for resource-efficient, performant "hard" multi-tenancy, particularly at the node level. An engineer from my company has a talk at the upcoming KubeCon on some experiences with Kata Containers. When we surveyed the state of the art in this area before doing that work, we found very very few people doing hard multi-tenancy at the node level with Kubernetes in production, even though various isolation technologies do exist (e.g. Kata, gVisor, Firecracker). It would be great to see progress in this area. Hopefully folks who are interested in this topic will engage with those project communities to push forward on making Kubernetes even more suited for hard multi-tenancy.

--

[Josh Gavant]

Thanks for your work @Tasha and group. Loved the guide you all published last year; and HNC is often requested at my customers. 🎉

TAG App Delivery is discussing multitenancy for cloud platforms and capabilities generally (i.e. beyond K8s) in CNCF's TAG App Delivery if folks want to participate: https://github.com/cncf/tag-app-delivery/issues/324

[Benjamin Elder]

Formal archival: https://github.com/kubernetes/community/pull/7362