What is actually happening
Microsoft says Data Sync will retire on 30 September 2027. Neowin's "a year from now" is a loose paraphrase of that date. The retirement has been public for some time. Microsoft announced the September 30, 2027 date in 2024.
The restriction on newcomers came earlier. From September 9, 2026, new Data Sync deployments can't be created in Azure subscriptions that have never used the service. The change affects only those subscriptions. The Register reports that subscriptions already using the service can keep creating, modifying and managing sync groups and member databases until retirement.
Neowin also says some functionality will stop working earlier than the final date, by design. Treat that as reported rather than detailed. The Microsoft pages I reviewed don't spell out which features or when. Don't assume your sync groups will run untouched until the last day.
Microsoft's public Azure update for the retirement says Data Sync workloads are deleted and their associated application data lost on retirement day. I couldn't confirm that wording independently here, so check it against your own sync group setup.
Why there is no drop-in replacement
Data Sync is a hub-and-spoke service. It lets you synchronize selected data bidirectionally across multiple databases, on premises and in the cloud. The hub must be an Azure SQL Database, and member databases can be Azure SQL Database or SQL Server. Microsoft's service documentation lists three main uses:
- Hybrid synchronization between SQL Server and Azure SQL Database.
- Reporting copies, which Microsoft calls distributed applications.
- Cross-region synchronization for globally distributed applications.
Microsoft's migration page names many alternatives. They include Azure Data Factory, Azure Functions, read replicas, linked servers, database mirroring, availability groups and transactional replication. The Register quotes Microsoft saying "there is no single replacement that maps to every SQL Data Sync configuration."
Analyst coverage is blunter. Directions on Microsoft says there are no automated migration tools and that customers should plan a significant replacement project. That is analyst opinion, not Microsoft guidance. Still, it fits what the matrix shows.
The requirements matrix, translated
Neowin reproduces Microsoft's matrix. The table below reflects its rows. Each row is Microsoft's judgment as relayed by Neowin, not a universal guarantee.
| If you need | Ruled out | Still in play |
|---|---|---|
| Bidirectional, both sides writable | Geo-replication, read replicas, standard transactional replication, Fabric mirroring | Redesign to a single writer, or a custom solution with Azure Data Factory or Azure Functions and your own conflict logic |
| Read-only copy in the same database | Anything heavier than needed | Read scale-out (read replicas) |
| Read-only copy in another region | Read scale-out | Active geo-replication, Azure Data Factory incremental copy |
| SQL Server to Azure SQL Database, near real time | Azure Data Factory batch schedules, if latency is tight | Transactional replication with Azure SQL Database as push subscriber |
| Azure SQL Database as the replication source | Transactional replication, because Azure SQL Database can't be a publisher | Azure Data Factory, Azure Functions, geo-replication |
| Subset of tables or columns, with transformation | Geo-replication and read replicas, which copy the whole database | Azure Data Factory, Azure Functions |
| Analytics in Fabric | Not listed | Fabric mirroring |
The traps worth knowing about
Replicas are not sync. Microsoft's warning, as quoted by Neowin, is that active geo-replication and read replicas give you a read-only copy of the whole database. You can't write to the secondary and you can't pick tables. If your Data Sync group moves a handful of tables, a replica copies far more than you need. It also won't accept writes on the other end.
Bidirectional is the hard case. Microsoft's stated position is that no managed service replaces it directly. Its advice is to check whether the application really needs to write in two places, since moving to a single writer is often simpler and more reliable. That's a recommendation, and it assumes you can change the application.
Data Sync gave you a built-in conflict policy. Microsoft's service documentation says the policy is a group-level setting, either Hub wins or Member wins. If you build two-way sync yourself with Data Factory or Functions, that logic becomes your code. Nothing I reviewed says those services reproduce Data Sync's behavior automatically.
Transactional replication has a direction. Microsoft's own comparison notes that you can't publish from Azure SQL Database. A source in Azure SQL Database therefore needs a different route.
Managed Instance was never covered. Microsoft's migration page notes that SQL Data Sync does not support Azure SQL Managed Instance. That limit shapes the source-and-destination grid on Microsoft's page. Check your exact source and target pairing before picking a tool.
Fabric mirroring is analytics-oriented. Microsoft lists mirrored databases from Azure SQL Database as an alternative. The matrix places it under analytics in Fabric, not under writable two-way sync.
Microsoft's recommendations in practice
Microsoft's own technical community posts lean toward Data Factory for Azure-to-Azure cases. One post describes Azure Data Factory with incremental changes (CDC) as the preferred path in a customer scenario because it is Azure-native, scalable and supported long term. Another asks whether you need two-way synchronization, or just a readable copy of the database somewhere else. That question will rescue a surprising number of migrations.
Microsoft's community guidance also describes running the new pipeline alongside Data Sync for a while and comparing results before cutting over. Staging a parallel run is a sensible approach whichever tool you pick.
A practical migration sequence
This sequence is my synthesis of Microsoft's guidance. It isn't an official procedure.
- Inventory every sync group. Record the hub, the members and the sync schema, which means the tables and columns involved.
- Classify each group. Note the sync direction, which endpoints accept writes, how much latency is acceptable, and whether the copy serves reporting, regional reads or application operations.
- Match to the matrix. Use the table above, and confirm the platform pairing on Microsoft's migration page.
- Challenge bidirectional requirements. Ask whether one writer would do.
- Pilot in parallel. Run the replacement beside Data Sync and compare the results before switching.
- Decommission cleanly. Microsoft's checklist, as Neowin relays it, says to delete the sync group and uninstall the Data Sync agent from on-premises SQL Server hosts. It also says to drop the
_dss_triggers and DataSync schema objects from every member, and to remove firewall rules and credentials that only Data Sync used.
Neowin says Microsoft's blog provides inventory queries. I haven't reproduced them, so use the originals if you want them. Check each cleanup step against your own deployment before running it.
A security reason not to wait
Microsoft's service documentation says Data Sync requires SQL authentication and doesn't support Microsoft Entra authentication. It adds that static passwords miss protections such as multifactor authentication, Conditional Access and managed identities. Microsoft recommends choosing replacements that support Entra authentication or managed identities. Data Sync also doesn't support network security perimeter. If your logical server sits behind a perimeter, that is another reason to move.
Bottom line
With about ten months left, the practical task is deciding which of several different tools fits each workload. Read-only, whole-database copies point to replicas or geo-replication. Selective or transformed movement points to Data Factory or Functions. One-way SQL Server to Azure feeds point to transactional replication. True two-way writes need a redesign or custom conflict handling that you own. Start the inventory now, because the hardest groups will take the longest.
References
- Microsoft offers alternatives for SQL Data Sync - Neowin Neowin · 2026-11-17T17:24:02+00:00
- Azure updates | Microsoft Azure azure.microsoft.com
- SQL Data Sync retirement migration - Azure SQL Database learn.microsoft.com