Skip to content

Scaling

Horizontal and Vertical Scaling Recommendations

Document Version: 1.0.0
Last Updated: 2026-02-10

1. Scaling Matrix

Component Horizontal Vertical Notes
Web Interface ✅ 1–10+ replicas ✅ Stateless; HPA recommended
API Server ⚠️ 1 replica* ✅ *Single replica per license; contact support for multi-replica
OCSP Responder ✅ 1–5+ replicas ✅ Read-only DB access, stateless
PostgreSQL ✅ Read replicas ✅ Primary + replicas for read scaling
Redis ✅ Sentinel/Cluster ✅ Optional HA deployment

2. Capacity Planning

Secrets Volume

Secrets Count API Server PostgreSQL Redis
< 10,000 2 CPU / 4 GB 2 CPU / 4 GB / 50 GB 1 CPU / 1 GB
10,000–50,000 4 CPU / 8 GB 4 CPU / 8 GB / 100 GB 2 CPU / 2 GB
50,000–200,000 4 CPU / 16 GB 8 CPU / 16 GB / 200 GB 2 CPU / 4 GB
200,000+ 8 CPU / 32 GB 16 CPU / 32 GB / 500 GB+ 4 CPU / 8 GB

Concurrent Users

Users Web Interface Replicas API Server Resources
< 50 1–2 2 CPU / 4 GB
50–200 2–3 4 CPU / 8 GB
200–500 3–5 4 CPU / 16 GB
500+ 5+ (HPA) 8 CPU / 32 GB

Agent Connections

Agents API Server Resources Notes
< 50 Base resources Standard heartbeat interval
50–200 +2 GB RAM Increase heartbeat interval to 120s
200–500 +4 GB RAM Consider connection pooling
500+ +8 GB RAM Contact support for tuning

3. Kubernetes HPA Configuration

Web Interface

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: mazevault-frontend-hpa
  namespace: mazevault
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: mazevault-frontend
  minReplicas: 2
  maxReplicas: 10
  metrics:
    - type: Resource
      resource:
        name: cpu
        target:
          type: Utilization
          averageUtilization: 70
    - type: Resource
      resource:
        name: memory
        target:
          type: Utilization
          averageUtilization: 80

OCSP Responder

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: mazevault-ocsp-hpa
  namespace: mazevault
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: mazevault-ocsp
  minReplicas: 1
  maxReplicas: 5
  metrics:
    - type: Resource
      resource:
        name: cpu
        target:
          type: Utilization
          averageUtilization: 70

4. Database Scaling

Read Replicas (Azure)

az postgres flexible-server replica create \
  --resource-group rg-mazevault-prod \
  --replica-name psql-mazevault-replica-1 \
  --source-server psql-mazevault-prod

Connection Pooling

For high-concurrency deployments, consider PgBouncer:

# pgbouncer.ini equivalent settings
pool_mode = transaction
max_client_conn = 400
default_pool_size = 25

5. Performance Tuning

PostgreSQL

Parameter Small (<10K secrets) Medium (10K–100K) Large (100K+)
shared_buffers 1 GB 4 GB 8 GB
effective_cache_size 3 GB 12 GB 24 GB
work_mem 16 MB 64 MB 128 MB
max_connections 100 200 400

Redis

Parameter Value Description
maxmemory 50-75% of available RAM Memory limit
maxmemory-policy allkeys-lru Eviction policy