Virtual network data gateway for Dynamics 365
How the virtual network (VNet) data gateway enables Dynamics 365 and Power Platform to access data in private Azure VNets — architecture, configuration, and the use cases.
Dynamics 365 and Power Platform often need to reach data sitting in private networks — Azure SQL Databases with private endpoints only, on-prem databases via ExpressRoute, services behind firewalls. Traditionally, this required the on-premises data gateway — a Windows service installed in the customer environment. The virtual network data gateway is a managed alternative: fully Microsoft-hosted, deployed into the customer's Azure virtual network, no Windows server to maintain.
The on-premises gateway recap.
- A Windows service installed in customer infrastructure.
- Polls Microsoft cloud for work.
- Bridges Power Platform / Dynamics requests to private data.
- Customer-managed — patching, scaling, monitoring.
- Available for over a decade; mature; widely used.
The virtual network data gateway.
- Managed by Microsoft — no Windows server to install or patch.
- Deployed into customer's Azure virtual network — a subnet hosts the gateway.
- Accesses resources inside the VNet — Azure SQL with private endpoint, Azure Function, etc.
- Connects to Microsoft cloud through standard mechanisms.
The headline benefit: customer's data stays inside their VNet; Microsoft's compute reaches in temporarily to fetch what's needed.
Architecture.
Power Platform / Dynamics 365
|
v
Microsoft cloud (gateway service)
|
v
Customer Azure VNet
subnet for gateway
|
v
Customer private resources
The gateway is provisioned in the customer's VNet but operated by Microsoft. The customer controls the VNet, the NSG rules, the resources it accesses; Microsoft operates the gateway runtime.
Configuration.
- Customer Azure subscription has a VNet with appropriate routing.
- A subnet is dedicated for the gateway.
- Subnet is delegated to
Microsoft.PowerPlatform/vnetaccesslinks. - In Power Platform admin, the gateway is created referencing the VNet/subnet.
- Data sources in the VNet are added to the gateway's connections.
- Power Apps, flows, and Power BI reports use the gateway to reach the private resources.
The setup is mostly Azure VNet plumbing plus a Power Platform reference.
Use cases.
- Azure SQL with private endpoint — public access disabled; gateway reaches via private endpoint.
- Azure Function with VNet integration — internal logic accessible to Power Automate.
- Private Azure Storage — files in storage accounts without public access.
- Azure Cosmos DB private — private endpoints.
- Custom Azure-hosted APIs — internal services accessible only inside VNet.
- On-prem via ExpressRoute — if VNet is connected to on-prem via ExpressRoute, gateway reaches on-prem resources too.
Comparison with on-premises gateway.
| Aspect | On-prem gateway | VNet gateway | |---|---|---| | Hosting | Customer Windows server | Microsoft managed | | Patching | Customer | Microsoft | | Scaling | Manual / replica | Auto | | HA | Cluster of installs | Built-in | | Data path | Through customer infrastructure | Through Azure | | Use case | On-prem databases | Azure VNet resources |
VNet gateway is the modern path for Azure-resident private resources; on-prem gateway remains for true on-prem data.
Authentication and authorisation.
- Power Platform / Dynamics user authenticates to Power Platform.
- Gateway authenticates to data source (SQL credentials, managed identity, OAuth).
- Identity flow depends on connector — some support pass-through, some require fixed credentials.
For SQL access, the cleanest pattern is managed identity for the gateway connecting to SQL.
Performance.
- Throughput — comparable to direct connections.
- Latency — slight added latency from cross-region or VNet routing.
- Connection pooling — gateway maintains connection pools.
For high-volume queries, gateway performance is reasonable; bulk data movement still benefits from ADF or other dedicated tools.
Networking specifics.
- NSG rules — control what the gateway subnet can reach.
- Private endpoints — point to specific resources without public routing.
- DNS — Azure Private DNS Zones for name resolution.
- Service endpoints — for some Azure services if needed.
VNet design matters; misconfigured networking gives confusing errors.
Monitoring.
- Connection health — visible in Power Platform admin.
- Query metrics — request counts, latency.
- Errors — connection failures logged.
The managed gateway has better observability than the on-prem alternative; Azure-native logs.
Common pitfalls.
- NSG too restrictive. Gateway can't reach target; cryptic errors.
- Wrong subnet delegation. Subnet not delegated to Power Platform service; gateway provisioning fails.
- Private endpoint DNS broken. Gateway resolves to public IP, doesn't find resource; private DNS zones needed.
- No high availability planning. Single gateway; outage affects all dependent flows.
- Connection limits. Many flows hitting the same gateway; throttling.
- Cost surprise. VNet gateway has hourly cost; budget.
Cost. VNet gateway is paid per gateway hour:
- Reasonable for moderate usage.
- Multiple gateways multiply cost.
- Compare to on-prem gateway (free, but Windows server cost).
For Azure-resident workloads, VNet gateway often wins on TCO once Windows server admin is included in on-prem gateway cost.
Migration from on-prem gateway.
- Add VNet gateway alongside.
- Reconfigure connections to use VNet gateway.
- Validate flows still work.
- Decommission on-prem gateway.
Migration is gradual; both can coexist during transition.
Strategic positioning. The VNet data gateway is Microsoft's path forward for Power Platform / Dynamics 365 private network access. For new deployments where data sits in Azure VNets, it's the default. For existing on-prem gateway deployments, evaluate moving to VNet gateway as Azure infrastructure modernises.
The benefit isn't transformative — connections work either way — but operational simplicity (no Windows servers, no patching, no clustering) compounds over time. For modern, Azure-centric data architectures, the VNet gateway removes a longstanding piece of customer-managed infrastructure. Worth adopting for new work; worth migrating to gradually for existing.
Related guides
- Azure Data Factory with Dynamics 365How to use Azure Data Factory for Dynamics 365 data integration — connectors, common patterns, performance tuning, and when ADF is the right tool vs alternatives.
- Azure Functions for Dynamics 365 integrationsHow to use Azure Functions to extend and integrate Dynamics 365 — patterns, authentication, lifecycle, performance, and the trade-offs vs Power Automate.
- Azure Service Bus integration with DataverseHow Dataverse publishes change events to Azure Service Bus — registration, message format, queues vs topics, and resilient consumer patterns.
- Azure Synapse Link for DataverseHow Synapse Link replicates Dataverse data to Azure Data Lake Storage continuously — architecture, configuration, query patterns, and the path forward as Microsoft Fabric Link emerges.
- Logic Apps Standard vs ConsumptionThe two Logic Apps hosting models — Standard (single-tenant) vs Consumption (multi-tenant) — and how to choose between them for Dynamics 365 integrations.