Best practices for cluster isolation in Azure Kubernetes Service (AKS)
As you manage clusters in Azure Kubernetes Service (AKS), you often need to isolate teams and workloads. AKS provides flexibility in how you can run multi-tenant clusters and isolate resources. To maximize your investment in Kubernetes, these multi-tenancy and isolation features should be understood and implemented.
This best practices article focuses on isolation for cluster operators. In this article, you learn how to:
- Plan for multi-tenant clusters and separation of resources
- Use logical or physical isolation in your AKS clusters
Design clusters for multi-tenancy
Kubernetes provides features that let you logically isolate teams and workloads in the same cluster. The goal should be to provide the least number of privileges, scoped to the resources each team needs. A Namespace in Kubernetes creates a logical isolation boundary. Additional kubernetes features and considerations for isolation and multi-tenancy include the following areas:
- Scheduling includes the use of basic features such as resource quotas and pod disruption budgets. For more information about these features, see Best practices for basic scheduler features in AKS.
- More advanced scheduler features include taints and tolerations, node selectors, and node and pod affinity or anti-affinity. For more information about these features, see Best practices for advanced scheduler features in AKS.
- Networking includes the use of network policies to control the flow of traffic in and out of pods.
- Authentication and authorization include the user of role-based access control (RBAC) and Azure Active Directory (AD) integration, pod identities, and secrets in Azure Key Vault. For more information about these features, see Best practices for authentication and authorization in AKS.
- Containers include pod security policies, pod security contexts, scanning images and runtimes for vulnerabilities. Also involves using App Armor or Seccomp (Secure Computing) to restrict container access to the underlying node.
Logically isolate clusters
Best practice guidance - Use logical isolation to separate teams and projects. Try to minimize the number of physical AKS clusters you deploy to isolate teams or applications.
With logical isolation, a single AKS cluster can be used for multiple workloads, teams, or environments. Kubernetes Namespaces form the logical isolation boundary for workloads and resources.
Logical separation of clusters usually provides a higher pod density than physically isolated clusters. There's less excess compute capacity that sits idle in the cluster. When combined with the Kubernetes cluster autoscaler, you can scale the number of nodes up or down to meet demands. This best practice approach to autoscaling lets you run only the number of nodes required and minimizes costs.
Kubernetes environments, in AKS or elsewhere, aren't completely safe for hostile multi-tenant usage. Additional security features such as Pod Security Policy and more fine-grained role-based access controls (RBAC) for nodes make exploits more difficult. However, for true security when running hostile multi-tenant workloads, a hypervisor is the only level of security that you should trust. The security domain for Kubernetes becomes the entire cluster, not an individual node. For these types of hostile multi-tenant workloads, you should use physically isolated clusters.
Physically isolate clusters
Best practice guidance - Minimize the use of physical isolation for each separate team or application deployment. Instead, use logical isolation, as discussed in the previous section.
A common approach to cluster isolation is to use physically separate AKS clusters. In this isolation model, teams or workloads are assigned their own AKS cluster. This approach often looks like the easiest way to isolate workloads or teams, but adds additional management and financial overhead. You now have to maintain these multiple clusters, and have to individually provide access and assign permissions. You're also billed for all the individual nodes.
Physically separate clusters usually have a low pod density. As each team or workload has their own AKS cluster, the cluster is often over-provisioned with compute resources. Often, a small number of pods is scheduled on those nodes. Unused capacity on the nodes can't be used for applications or services in development by other teams. These excess resources contribute to the additional costs in physically separate clusters.
This article focused on cluster isolation. For more information about cluster operations in AKS, see the following best practices: