The following are the compute requirements for worker nodes in the Kubernetes cluster. We recommend three replicas per service for a “Production” environment except for the data stores (MySQL, MongoDB, Redis). For a “Staging” environment, we recommend one replica per service.
Production Environment:
For Customers Using In-Cluster Services (MySQL, MongoDB, MinIO):
Number of Worker Nodes: 5
CPU per Node: 4 vCPUs
Memory per Node: 16Gi
Default Storage per Node: 50GB minimum
Persistent Storage for Application Data: 500GB (allocated externally, not on the worker nodes)
For Customers Using External Datastores (S3, DocumentDB/Mongo Atlas, RDS MySQL):
Number of Worker Nodes: 3
CPU per Node: 4 vCPUs
Memory per Node: 16Gi
Default Storage per Node: 50GB minimum
Persistent Storage: Not required (handled externally)
Staging Environment:
For Customers Using In-Cluster Services (MySQL, MongoDB, MinIO):
Number of Worker Nodes: 2
CPU per Node: 4 vCPUs
Memory per Node: 16Gi
Default Storage per Node: 50GB minimum
Persistent Storage for Application Data: 500GB (allocated externally, not on the worker node)
For Customers Using External Datastores (S3, DocumentDB, Mongo Atlas, RDS MySQL):
Number of Worker Nodes: 2
CPU per Node: 4 vCPUs
Memory per Node: 16Gi
Default Storage per Node: 50GB minimum
Persistent Storage: Not required (handled externally)
Caveat on Storage:
Please ensure that each worker node is allocated at least 50GB of storage. Insufficient storage allocation (e.g., 20GB) will cause issues with larger container images, leading to deployment failures. Ensuring proper storage allocation is critical for smooth operations.