Azure Synapse: Spark 3.4 End-of-Life Pushes Migration to 3.5 and Microsoft Fabric
microsoft microsoft-fabric azure-synapse apache-spark data-integration data-migration
trending_up Trend: microsoft

Azure Synapse: Spark 3.4 End-of-Life Pushes Migration to 3.5 and Microsoft Fabric

calendar_month June 4, 2026 update Updated: June 4, 2026

Summary

The Azure Synapse Runtime for Apache Spark 3.4 reached its End-of-Life (EOL) on March 31, 2026. Microsoft is now actively pushing users to migrate to the more stable Spark 3.5 version. Simultaneously, the transition from classic Azure Synapse and Data Factory services to the new all-in-one analytics platform, Microsoft Fabric, is being emphasized as the strategic path forward.

What happened?

As of March 31, 2026, the Spark 3.4 runtime in Azure Synapse has been officially retired. This means that new Spark pools can no longer be created using this version, and existing workflows still running on Spark 3.4 are at risk. Microsoft recommends an in-place upgrade to Spark 3.5, which requires manual configuration adjustments within Synapse workspaces. In parallel, Microsoft is promoting Fabric as the “Next-Gen Analytics Leap” to migrate Synapse users to the new platform long-term.

Why it matters

For data engineers and businesses, this signifies an immediate need for action. Outdated runtimes can lead to security risks, incompatibilities, and ultimately the halt of business-critical data pipelines. Furthermore, the development clearly shows that Microsoft is scaling back investments in Synapse and focusing entirely on Microsoft Fabric. An early switch or at least a migration strategy is therefore essential.

Evidence

Official Microsoft documentation and reports from technology partners like Stoneridge Software confirm March 31, 2026, as the deadline. Following this EOS (End of Support) date, no security updates or bug fixes will be provided for Spark 3.4. In the Microsoft Fabric Community, the transition is communicated as a necessary step for modern analytical workloads.

Analysis

The migration from Spark 3.4 to 3.5 is technically mostly straightforward, as Spark 3.5 represents an evolutionary development. However, the process is a symptom of Microsoft’s larger consolidation strategy. By increasing migration pressure in Synapse, Microsoft is simultaneously lowering the barrier for switching to Fabric. Fabric offers an integrated environment combining Synapse, ADF, and Power BI, which aims to reduce complexity but also implies stronger vendor lock-in within the Microsoft ecosystem.

Practical Takeaways

  • Immediate Check: Audit all Synapse workspaces for active Spark 3.4 pools.
  • Upgrade Path: Create new Spark pools with version 3.5 and test existing notebooks and jobs against them.
  • Configuration Update: Switch Synapse Link and other integrations to the new pool ID.
  • Strategic Planning: Evaluate moving to Microsoft Fabric to benefit from improved integration and analytics features.

Open Questions

  • How long will Spark 3.5 be supported in Synapse before Fabric becomes the only option?
  • Are there specific libraries or UDFs (User Defined Functions) that might cause issues when moving to 3.5?

Sources

  1. Apache Spark 3.4 End of Support: How to Prepare for the 3.5 Upgrade
  2. From Azure Synapse and Azure Data Factory to Microsoft Fabric