Deployment Models

Datafi supports four deployment models to match your organization's infrastructure, compliance, and operational requirements. Each model offers a different balance of management overhead, data residency control, and network topology.

Overview

Comparison Table

Feature	SaaS	Private Cloud	On-Premises	Hybrid
Coordinator hosted by	Datafi	Customer	Customer	Datafi
Edge server hosted by	Datafi	Customer	Customer	Customer
Infrastructure management	Datafi	Customer	Customer	Shared
Data leaves your network	Yes	No	No	No
Auto-updates	Yes	Optional	Manual	Coordinator only
Setup complexity	Low	Medium	High	Medium
Best for	Fast start	Cloud-native regulated	Air-gapped / strict compliance	Data residency with managed control plane

SaaS (Fully Managed)

In the SaaS model, Datafi manages all infrastructure. The coordinator and edge servers run in Datafi's cloud, and your data sources connect over encrypted channels.

When to choose SaaS:

• You want the fastest path to production.
• Your data sources are accessible over the internet or via VPN peering.
• You do not have strict data residency requirements that prevent data from leaving your network.

Architecture:

Responsibilities:

Area	Datafi	Customer
Infrastructure provisioning	Yes	No
Software updates	Yes	No
Monitoring and uptime	Yes	No
Data source credentials	No	Yes
Access policies	No	Yes

Private Cloud (Customer's Cloud)

In the Private Cloud model, you deploy both the coordinator and edge servers in your own cloud account (AWS, Azure, or GCP). Datafi provides container images and deployment guides.

When to choose Private Cloud:

• You need full control over the infrastructure.
• Regulatory requirements mandate that data and compute remain within your cloud boundary.
• You have a platform engineering team to manage deployments.

Architecture:

On-Premises (Behind Firewall)

In the On-Premises model, all components run within your data center, behind your firewall. No traffic leaves your network.

When to choose On-Premises:

• You operate in an air-gapped or highly restricted network environment.
• Regulatory or policy constraints prohibit any cloud connectivity.
• You have the infrastructure team to manage bare-metal or VM-based deployments.

Architecture:

warning

On-premises deployments require manual software updates. Ensure you have a process to apply updates regularly to receive security patches and new features.

Hybrid (Coordinator Cloud + Edge On-Prem)

The Hybrid model places the coordinator in Datafi's cloud while your edge servers run on your network, close to your data sources. Your data never leaves your network -- only metadata and query plans are exchanged with the coordinator.

When to choose Hybrid:

• You want the operational simplicity of a managed coordinator but cannot allow raw data to leave your network.
• Your data sources are behind a firewall or in a private network.
• You need the coordinator's auto-update and monitoring capabilities without managing that infrastructure yourself.

Architecture:

What crosses the network boundary:

Data Type	Crosses Boundary	Direction
Query plans	Yes	Coordinator to Edge
Schema metadata	Yes	Edge to Coordinator
Raw query data	No	Stays within your network
Query results	No	Stays within your network
Credentials	No	Stays within your network

tip

The Hybrid model is the most popular choice for organizations that need data residency guarantees while minimizing infrastructure management overhead.

Choosing a Model

Use the following decision tree to select the right deployment model:

Next Steps

• Container Deployment -- learn how to deploy Datafi using Docker and Kubernetes.
• Azure Deployment Guide -- step-by-step Azure deployment with Container Instances, Redis, and Blob Storage.
• Operations Guide -- health monitoring, graceful shutdown, and rolling updates.