This article contains information about Essential PKS Support Matrix. This offering is no longer available after March 10, 2020.
This matrix is only delivered via the VMware Customer Reliability Engineering team.

VMware Tanzu Kubernetes Grid 1.x

VMware support for Kubernetes

• SEV 1 support for current minor version and two minor versions prior (n, n-1, n-2) (Inclusive of all major stable features)
• Upstream advocacy for features and defects
• On-demand, downloadable signed binaries
• Patching of critical issues prior to upstream releases

Prerequisites for Support

Choose the technologies from within each category below. These technologies are required to get started with Kubernetes.
Prerequisites technologies themselves are not supported through Essential PKS unless they are VMware products or Kubernetes components.

Infrastructure Platform

• AWS, AWS GovCloud (excluding China Cloud)
• Azure (excluding GovCloud and China Cloud)
• Bare metal
• vSphere beginning with v6.5
• GCP (excluding GovCloud and China Cloud)
• VMware Integrated OpenStack beginning with v6.0
• IBM Cloud (excluding GovCloud and China Cloud)
• VMC
• RedHat OpenStack Platform or by reviewed exception only
• Alibaba Cloud

Storage

• Any drivers–deemed stable by the vendor–designed to integrate with the Container Storage Interface
• K8s in-tree storage providers: iSCSI, NFS, AWS, vSphere, GCP, Azure

​Bootstrap (non-managed Kubernetes only)

• kubeadm (built-in to Kubernetes)

Kubernetes Node OS

• RHEL beginning with 7.4
• CentOS beginning with 7.2
• Ubuntu LTS beginning with 16.04
• Amazon Linux 2
• Photon OS (not for use on bare metal)

Low-level CNI (optional)

• Multus

Tooling Decisions

VMware requires choosing from the tools below and provide advisory support for them.
VMware OSS projects including Photon OS, Velero, Contour, Harbor (beginning with version 1.8.1) and Sonobuoy have break-fix support with subscription for most recent stable version.

​Container Runtime

• Docker-CE/Docker-EE (any versions validated with chosen Kubernetes versions)
• Containerd (versions validated with chosen Kubernetes versions) Only valid for Kubernetes versions beginning with 1.12
• CRI-O versions beginning with 1.14.5
• Singularity versions beginning with 3.3.0
  If you intend to use more than one type of OS and/or CRI in one K8s clusters, please consult with VMware Support before doing so to ensure important aspects have been considered.

Infrastructure Automation

• Any tool instrumenting kubeadm
  VMware only supports kubeadm, not the automation tool

Container Network

• Calico versions beginning with 2.6
• Canal versions beginning with 2.6
• NSX Container plugin beginning with v2.2
• Cilium beginning with v1.5.0 (only if used with a supported OS that supports Cilium with its default kernel)
• Weave versions beginning with 2.5.2

Ingress

• Contour (always use the latest stable versions) Your subscription includes break-fix support for this VMware OSS project
• Nginx (always use the latest stable version)
• Ambassador versions beginning with 0.50.0
• Voyager beginning with v10.0
• Istio beginning with v1.1

Service Mesh

• Istio beginning with v1.1
• NSX-Service Mesh (full support via an NSX-Service Mesh subscription)

Observability

• Prometheus (latest stable version)
• Grafana (latest stable version)
• Alert Manager (latest stable version)
• FluentD (latest stable version)
• Zipkin (latest stable version)
• Jaeger (latest stable version)

CI/CD

• Helm v2/v3

Architectural Recommendations

For environments that require high availability, VMware recommends the following, though it is not required to maintain supportability.

• Single-region cluster  (If cloud, the control plane is distributed in multiple Availability Zones)
• Standalone multi-node etcd cluster ​
• Highly available control plane​
• Use kubernetes datastore for CNI storage ​
• Use load balancer for kube-apiservers

 

Kubernetes Beta Features

We support only K8s beta features in which the feature gate has been removed. Otherwise, support for experimental features is by reviewed exception only. If you have a feature you'd like reviewed, please contact VMware support.

Workloads

We support both stateful / stateless workloads as it pertains to Kubernetes. While we are agnostic to the way in which those workloads are implemented and run within the pod, our guidance with the various technologies used at that level of the stack will be minimal. 

Projects within Heptio Labs are considered experimental, thus no support is provided.

FAQ:

Q: What "break-fix support" mean?
A: Break-fix support means that we have VMware engineers resourced to work directly upstream to either a) fix issues directly or b) partner with the community on a fix.

Q: What does does "advisory support" mean?
A: Advisory support means that we can provide support for the tool insofar as it is used within the context of Kubernetes. That said, we do not have VMware engineers assigned to those upstream projects and thusly cannot extend break-fix support.

Q: What is the difference between "Production Support" and "Break-Fix Support?"
A: Production Support is about the SLAs given a severity level and Break-fix is the degree to which we have control to fix the root cause

Tanzu Kubernetes Grid Plus Support Matrix (78173)