Cluster specifications and sizing guide for Monobase Infrastructure.
For single client with core components only:
Nodes: 3 worker nodes CPU: 4 cores per node (12 total) Memory: 16GB per node (48GB total) Storage: 100GB per node (for system + Longhorn)
Workload:
- Monobase API (2 replicas)
- Monobase Account (2 replicas)
- PostgreSQL (3 replicas)
- Longhorn, Gateway, ArgoCD, External Secrets
Total Resource Usage:
- ~7 CPU cores
- ~23Gi memory
- ~100Gi storage (PostgreSQL PVC)
For single client with all optional components:
Nodes: 5 worker nodes CPU: 8 cores per node (40 total) Memory: 32GB per node (160GB total) Storage: 500GB per node (for Longhorn pool)
Workload:
- All core components
- API Worker (2 replicas)
- MinIO (6 replicas)
- Valkey (3 replicas)
- Monitoring stack
Total Resource Usage:
- ~22 CPU cores
- ~53Gi memory
- ~1.15TB storage
For multiple clients on same cluster:
Formula:
- Base infrastructure: 3 CPU, 7Gi memory (shared)
- Per client (core): +7 CPU, +23Gi memory, +100Gi storage
- Per client (full): +22 CPU, +53Gi memory, +1.15TB storage
Example (5 clients, core stack):
- Infrastructure: 3 CPU, 7Gi
- 5 × clients: 35 CPU, 115Gi, 500Gi
- Total: 38 CPU, 122Gi, 500Gi
- Cluster: 10 nodes × 4 CPU × 16GB
Minimum: 1.25+ Recommended: 1.27+ Features required:
- Gateway API support (1.25+)
- Pod Security Standards (1.25+)
- CSI volume expansion (1.24+)
- Node Type: m6i.xlarge (4 vCPU, 16GB)
- Storage: gp3 EBS volumes
- Networking: VPC with 3 AZs
- IAM: IRSA for pod permissions
- Addons: EBS CSI driver, VPC CNI
- Node Type: Standard_D4s_v3 (4 vCPU, 16GB)
- Storage: Premium SSD
- Networking: Azure CNI
- Identity: Workload Identity
- Features: Azure Monitor integration
- Node Type: n2-standard-4 (4 vCPU, 16GB)
- Storage: pd-ssd
- Networking: VPC-native
- Identity: Workload Identity
- Features: GKE Autopilot (optional)
- Minimum: 100GB per node
- Recommended: 500GB per node
- Type: SSD preferred
- Replicas: 3x replication (3 nodes minimum)
- Small: 20-50Gi (<10k records)
- Medium: 50-200Gi (10k-100k records)
- Large: 200Gi-1Ti (100k+ records)
- Growth: Plan for 2x per year
- 1TB usable: 6 nodes × 250Gi = 1.5TB raw (EC:2)
- 2TB usable: 6 nodes × 500Gi = 3TB raw (EC:2)
- Or use external S3 for >1TB
Bandwidth:
- Minimum: 1 Gbps between nodes
- Recommended: 10 Gbps for production
- Internet: 100 Mbps minimum
Latency:
- Between nodes: <10ms (same AZ preferred)
- To internet: <100ms
Ports:
- 443 (HTTPS) - LoadBalancer
- 6443 (K8s API) - Control plane
- Internal cluster networking
- Dedicated VPC/VNet (isolated network)
- Encryption at rest (EBS/disk encryption)
- Encryption in transit (TLS)
- Audit logging enabled
- Appropriate agreements with cloud provider (BAA for HIPAA, DPA for GDPR, etc.)
- Access controls (IAM/RBAC)
- Private nodes (no public IPs)
- Bastion host for access
- Network policies enabled
- Pod Security Standards enforced
- Secrets in KMS
For more details, see HIPAA-COMPLIANCE.md (covers HIPAA and general compliance).