A single Kubernetes cluster provides an extremely flexible platform to design any complex distributed system. This also includes statefull workloads such as databases.
For many scenarios a single cluster is enough.
- scaleability achieved by increasing number of nodes from a node pool
Kubernetes 1.24 supports clusters up to 5000 nodes
- workload isolation using namespaces
When are multiple clusters needed.
- there are already existing clusters that need to talk to each other
- strict security and workload isolation rules
- cost factors