Migrating Azure SQL Database to Nutanix Database Service with a Planned Cutover Window

In today’s multi-cloud landscape, flexibility and data portability are increasingly important considerations. While public cloud database services offer convenience, some organizations may reassess public-cloud database deployments based on data-transfer charges, scaling requirements, portability, and operating-model considerations.

To regain control over their data, optimize total cost of ownership (TCO), and achieve consistent performance, some organizations evaluate private or hybrid-cloud environments for selected workloads.

When it comes to database infrastructure, NDB provides a Database-as-a-Service (DBaaS) operating model for database lifecycle management across supported environments.

If you are currently planning to migrate an Azure SQL database back to an SQL Server instance deployed and managed through NDB, this guide provides a comprehensive architectural roadmap and execution framework to complete the journey with a short planned outage, subject to validation.

Why Nutanix Database Service (NDB)?

NDB simplifies database lifecycle management across hybrid environments. Key advantages include:

  • Operational Efficiency: NDB can automate selected database lifecycle-management tasks. NDB automates patching, provisioning, and complex backup workflows.
  • Cost Optimization: By consolidating traditional, siloed infrastructure, NDB provides opportunities to improve infrastructure and operational cost predictability.
  • Support for Multiple Database Engines: Manage SQL Server, Oracle, MySQL, PostgreSQL, MariaDB, and MongoDB through a centralized management experience.

Choosing Your Migration Path: A Quick-Reference Matrix

Before diving into configuration, it is critical to select the right tool for your specific database architecture. Depending on your table structures, data volume, and network constraints, you will typically leverage either Azure SQL Data Sync or Azure Data Factory (ADF).

Evaluation MetricMethod 1: Azure SQL Data SyncMethod 2: Azure Data Factory (ADF)
Ideal ForSmall-to-medium OLTP databases.ADF may be appropriate for large or complex migrations requiring pipeline orchestration, subject to throughput and supportability validation.
Primary Key RequirementStrictly Mandatory for all synchronized tables (confirm from Microsoft documentation)Not Mandatory (Flexible schema support, Primary keys are not inherently required for a basic copy, but the incremental design still requires a reliable change-detection and reconciliation strategy).
Synchronization LogicContinuous, one-way cloud-to-local replication group.Incremental "delta" copies triggered by watermark columns.
Setup ComplexityLower (Relies on a lightweight local agent).Higher (Requires pipelines, links, and integration runtimes).
Scale LimitationsBest suited for moderate data volumes and standard data types.Engineered for high-volume, enterprise-grade data movement.
DowntimeBased on the synch frequencyBased on the ADF synch interval

Pre-Migration Checklist & Performance Optimizations

To help prepare for a controlled data transfer, ensure the following foundational prerequisites are met before initiating either migration method:

  • Network and Firewall Topography: Verify that secure network routing is established and validate the required source, agent or integration-runtime, target, direction, protocol, DNS, TLS, proxy, and port requirements for the selected migration architecture.
  • Target Environment Readiness: Your target SQL Server VM (e.g., NDB-SQL-VM1) should already be provisioned via NDB, containing a blank target database (e.g., DB1) ready to receive data.
  • Schema Deployment: Before moving the data payload, generate your full database object scripts or a DACPAC from Azure SQL and deploy the schema directly to your target NDB VM.
  • Maximizing Throughput: To prevent performance bottlenecks during the initial bulk data load, review the NDB database-protection policy before the initial load. If reducing protection overhead is considered, do so only under an approved change procedure after confirming recovery-point, retention, and rollback requirements. This option may reduce protection overhead during the initial load, but its effect must be validated in the target environment when populating the new database.

Method 1: Azure SQL Data Sync

Azure SQL Data Sync is ideal for databases where all tables have Primary Keys. It allows for one-way synchronization from the cloud "Hub" to the NDB "Member."

High-Level Architecture

Architecture of the Sync Group showing the Hub (Azure SQL) and Member (NDB SQL VM) relationship.

Figure 1: Architecture of the Sync Group showing the Hub (Azure SQL) and Member (NDB SQL VM) relationship.

Key Migration Steps:

  1. Our synchronization pipeline begins directly inside the Azure Portal. First, we establish a new Sync Group, designating our source Azure SQL database as the central Hub.
Defining the central Sync Group parameters, hub credentials, and the desired refresh frequency within the Azure Portal.

Figure 2: Defining the central Sync Group parameters, hub credentials, and the desired refresh frequency within the Azure Portal.

  1. Next, we deploy the synchronization agent to facilitate secure communication with Azure by deploying the lightweight Data Sync Agent onto the target NDB VM. This agent creates an outbound connection from the agent to the required Azure services, subject to the applicable network configuration with Azure without requiring wide-open inbound firewall rules.
Registering the target NDB environment inside the on-premises Data Sync Agent utility.

Figure 3: Registering the target NDB environment inside the on-premises Data Sync Agent utility.

  1. With the agent online, we jump back to the cloud console to register our NDB database as an official Sync Member. During this step, ensure that the data flow direction is explicitly set to "From the Hub". Finally, we select the specific tables targeted for migration.
Finalizing member rules, setting traffic direction to pull from the cloud, and mapping schema tables.
Finalizing member rules, setting traffic direction to pull from the cloud, and mapping schema tables.

Figure 4: Finalizing member rules, setting traffic direction to pull from the cloud, and mapping schema tables.

Strategic Limitations & Risk Mitigations

While Data Sync is highly effective, enterprise environments must account for several technical guardrails:

  • The Primary Key Constraint: Every table included in the sync group must have an explicit Primary Key. If your legacy schema includes heap tables, you must introduce surrogate (IDENTITY) keys before starting the process.
  • Unsupported Data Types: Certain specialized schemas, including computed columns and specific user-defined types, cannot be automatically replicated through this method and require separate scripting.
  • Trigger Conflicts: Azure SQL Data Sync relies on its own internal tracking triggers. Review your existing Data Manipulation Language (DML) triggers to ensure they don't conflict with data ingestion.
  • Post-Migration Identity Reseeding: Because identity values are synced across distinct systems, you must manually validate and reseed identity columns post-migration to prevent duplicate key generation errors when applications resume write operations.

Method 2: Azure Data Factory (ADF)

For massive, enterprise-scale data volumes or complex schemas that lack uniform primary keys, Azure Data Factory (ADF) offers a robust pipeline-driven approach. ADF can use a watermark or another documented change-detection strategy; the design must explicitly address inserts, updates, deletes, retries, and watermark persistence and sequentially sync delta records down to NDB provisioned on-prem databases.

High-Level Architecture

Enterprise data plumbing showcasing the network communication path between Azure Data Factory and the NDB target environment via a Self-hosted Integration Runtime.

Figure 5: Enterprise data plumbing showcasing the network communication path between Azure Data Factory and the NDB target environment via a Self-hosted Integration Runtime.

Key Migration Steps:

  1. To kick off an ADF-driven migration, we install a Self-hosted Integration Runtime (IR) directly onto a VM within the target Nutanix network topology. This Self-hosted Integration Runtime serves as the local processing bridge for pipeline execution, securely pulling down data payloads dictated by the cloud pipeline.
Verifying the operational status of the on-premises Self-hosted Integration Runtime within the ADF dashboard.

Figure 6: Verifying the operational status of the on-premises Self-hosted Integration Runtime within the ADF dashboard.

  1. With our compute bridge online, we define our Linked Services and Datasets. These act as the explicit connection strings mapping our Azure SQL source to our local NDB target. Leveraging dynamic content here allows us to loop through multi-table databases cleanly.
Mapping connection endpoints for both the source cloud repository and the destination database engine.

Figure 7: Mapping connection endpoints for both the source cloud repository and the destination database engine.

  1. Finally, we construct and initiate our incremental migration pipeline. The architecture executes lookup operations to determine the latest "watermark" value, copies the delta data, and stores the new watermark baseline for subsequent iterations.
Real-time monitoring of the iterative pipeline execution, validating data movement and watermark precision.

Figure 8: Real-time monitoring of the iterative pipeline execution, validating data movement and watermark precision.

Strategic Limitations & Risk Mitigations

Migrating at scale via ADF requires strict adherence to schema best practices:

  • Watermark Dependency: Delta loading hinges on having a highly reliable, incrementing column (such as a LastModifiedDate or RowVersion timestamp) across your tables.
  • Foreign Key Serialization: When copying multi-table environments simultaneously, data loads can fail if foreign key dependencies are violated during ingestion. Ensure your pipeline structures copy parent tables prior to child data.
  • Network Cost Overhead: Because ADF actively pulls raw datasets out of the public cloud, architects should monitor data egress volumes to anticipate variable cloud provider network billing.

Final Cutover and Data-Integrity Validation Checklist

Once your data is successfully replicating and your delta gap is closed, it is time to execute your final production cutover window. Follow these steps to support data-integrity validation:

Step 1: Quiesce and Stop Traffic

Pause all inbound application transactions routing to the live Azure SQL database. This freezes the source state and prevents mid-cutover data discrepancies.

Step 2: Run the Final Sync

Trigger a final manual sync iteration via Data Sync or your ADF pipeline.The final synchronization duration depends on remaining change volume, transaction rate, network capacity, synchronization latency, and pipeline performance. Measure it during a rehearsal to understand the entirety of your application's actual data-downtime window.

Step 3: Run Technical Integrity Checklists

Before opening up connection routes to the new NDB instance, execute these verification commands on the target SQL engine:

  • Data Consistency Check: Run row-count and checksum scripts across your critical tables to identify discrepancies between source and destination.
  • Database Structural Integrity: Run DBCC CHECKDB to check for structural issues during bulk parsing.
  • Identity Value Alignment: Execute DBCC CHECKIDENT on your tables to ensure the identity seeds are correctly configured, avoiding duplicate primary key exceptions on subsequent app writes.

Step 4: Redirect Application Routes

Update your core application connection strings to target your SQL Server instance managed through NDB. Restore user traffic, check performance baselines, and re-enable your NDB Time Machine protection rules to begin automated snapshots.

Conclusion

Migrating your primary databases off public cloud instances doesn't have to imply massive disruption or daunting project timelines. Whether you opt for the continuous, straightforward synchronization of Azure SQL Data Sync or the enterprise-scale orchestration capabilities of Azure Data Factory, Nutanix Database Service (NDB) serves as a robust target platform for enterprise SQL workloads.

Migrating to NDB can provide greater control over the target deployment and operating model, deliver infrastructure and operational cost predictability, and reduce dependence on a single public-cloud database service, all while retaining a DBaaS-oriented operating model for supported database operations.

©2026 Nutanix, Inc. All rights reserved. Nutanix, the Nutanix logo and all Nutanix product and service names mentioned are registered trademarks or trademarks of Nutanix, Inc. in the United States and other countries. All other brand names mentioned are for identification purposes only and may be the trademarks of their respective holder(s).