Infrastructure Sizing

Use the sizing calculator below to determine the right infrastructure for your Datafly Signal deployment. Select your cloud provider, expected monthly event volume, number of destinations, and average event size to get recommended specs and an itemised cost estimate.

Cloud Provider

Region: EU (Hetzner DE / IONOS DE / DigitalOcean AMS)

Monthly Events

Destinations / Integrations

Avg Event Size

$9.98

Estimated monthly cost

Starter

Recommended tier

~12/sec

Peak capacity (3x avg)

19.1 GB in / 57.2 GB out

Monthly data transfer

VPS not recommended for this volume

For 10M events/month you need Kubernetes with managed services. Switch to AWS, GCP, or Azure above for accurate pricing. The costs shown below are based on VPS pricing and would not provide the required capacity.

Compute Nodes

Cost Breakdown

Component	Specification	Monthly Cost	% of Total
Compute Nodes	2x Hetzner CX22 (2 vCPU, 4 GB)	$9.98	100.0%
Total Estimated Monthly Cost		$9.98	100%

Prices are on-demand rates for EU (Hetzner DE / IONOS DE / DigitalOcean AMS). Reserved instances and committed use discounts can reduce costs by 30-60%. Data transfer costs assume internet egress — VPC-internal traffic is free or cheaper. Last updated March 2026.

Helm Values Snippet

Copy this into your values.yaml as a starting point. See the VPS deployment guide for the complete configuration.

# Datafly Signal — VPS Starter Tier
# 10M events/month, 3 destinations
# ~12 peak events/sec
# Estimated cost: ~$9.98/month

ingestionGateway:
  replicas: 1
  hpa:
    enabled: false
    minReplicas: 1
    maxReplicas: 4

eventProcessor:
  replicas: 1
  hpa:
    enabled: false
    minReplicas: 1
    maxReplicas: 4

deliveryWorkers:
  vendors:
    - name: webhook
      enabled: true
      replicas: 1
      hpa:
        enabled: false
        minReplicas: 1
        maxReplicas: 4

identityHub:
  replicas: 1

managementApi:
  replicas: 1

managementUi:
  replicas: 1

Sizing Tiers

Datafly Signal scales across five sizing tiers based on monthly event volume. Each tier specifies Kubernetes node requirements, Kafka broker capacity, Redis and PostgreSQL instance sizes, and estimated monthly cloud costs.

Tier	Monthly Events	Peak Events/sec	Infrastructure	Estimated Cost
Micro	≤ 5M	≤ 5	1 VM/VPS, Docker Compose	~$5-35/mo
Starter	≤ 10M	≤ 10	2 K8s nodes, 1 broker	~$300/mo
Small	≤ 100M	≤ 50	3 K8s nodes, 2 brokers	~$900/mo
Medium	≤ 1B	≤ 400	3 K8s nodes, 3 brokers	~$1,800/mo
Large	≤ 10B	≤ 4,000	5 K8s nodes, 5 brokers	~$4,500/mo
XL	≤ 100B	≤ 40,000	8 K8s nodes, 7 brokers	~$12,000/mo

Cost estimates include all infrastructure: Kubernetes nodes, managed Kafka, Redis, PostgreSQL, load balancer, data transfer (ingress and egress), storage, and DNS. Actual costs vary by reserved instance pricing and committed use discounts (30-60% savings typical).

Service Replica Counts by Tier

Service	Starter	Small	Medium	Large	XL
Ingestion Gateway	1	2	3	5	8
Event Processor	1	2	3	5	8
Delivery Workers	1	2	3	5	10
Identity Hub	1	1	2	2	3
Management API	1	2	2	2	3
Management UI	1	2	2	2	2

Kafka Partition Recommendations

Kafka partition counts should match or exceed the maximum replica count of consumers for each topic to ensure parallel processing:

Topic	Starter	Small	Medium	Large	XL
`events.raw`	3	6	12	24	48
`events.processed`	3	6	12	24	48
`events.delivery.*`	3	6	12	24	48
`events.dlq`	1	3	6	12	24

⚠️

Kafka partition counts can be increased but not decreased. Start with the recommended count for your tier and increase as needed.

PostgreSQL Storage

PostgreSQL storage requirements depend on the number of organisations, sources, integrations, and audit log retention:

Tier	Recommended Storage	IOPS
Starter	10 GB	3,000
Small	20 GB	3,000
Medium	50 GB	3,000
Large	100 GB	6,000
XL	200 GB	12,000

Horizontal Pod Autoscaling

For production deployments, enable Horizontal Pod Autoscaling (HPA) to automatically scale services based on CPU utilisation:

ingestionGateway:
  hpa:
    enabled: true
    minReplicas: 2
    maxReplicas: 10
    targetCPUUtilizationPercentage: 70
 
eventProcessor:
  hpa:
    enabled: true
    minReplicas: 2
    maxReplicas: 10
    targetCPUUtilizationPercentage: 70

The sizing tiers above define minReplicas. HPA will scale up to maxReplicas during traffic spikes and scale back down during quiet periods.

Next Steps

Choose your cloud provider and follow the deployment guide: AWS, GCP, or Azure
Review the Kubernetes page for Helm chart details
Set up Observability for monitoring and alerting

Customer-Hosted AWS