Amazon RDS for SQL Server now supports Microsoft SQL Server 2025
SQL Server 2025 on RDS adds native REST-from-T-SQL, vector data types, and expanded Standard Edition limits—enabling AI workloads without re-architecting apps.
View original announcement →Visual Summary
graph TD
A{{RDS SQL Server 2025}}:::announced
B(Amazon Bedrock):::compute
C(AWS Lambda):::compute
D(Amazon S3):::storage
E(Amazon SageMaker):::compute
F([Native REST from T-SQL]):::feature
G([Native Vector Data Type]):::feature
H([Resource Governor]):::feature
I((On-Premises SQL Server)):::external
J((Application Users)):::external
J ==>|"queries"| A
A -->|"invokes"| F
F -->|"calls"| B
F -->|"triggers"| C
F -->|"reads/writes"| D
F -.->|"infers"| E
A -->|"stores embeddings"| G
A -->|"manages workloads"| H
I -.->|"migrates to"| A
classDef announced fill:#ff9900,stroke:#ec7211,color:#fff,font-weight:bold
classDef compute fill:#e3f2fd,stroke:#1565c0,color:#1565c0
classDef storage fill:#e8f5e9,stroke:#2e7d32,color:#2e7d32
classDef feature fill:#fff3e0,stroke:#e65100,color:#e65100
classDef external fill:#f5f5f5,stroke:#616161,color:#616161
What's New
Amazon RDS for SQL Server now supports Microsoft SQL Server 2025 across Enterprise, Standard, and Developer editions, bringing AI-native database capabilities to the fully managed RDS service. SQL Server 2025 introduces native REST endpoint invocation from T-SQL, a native vector data type, and significant Standard Edition capacity increases, all available without re-architecting existing applications. Customers can upgrade existing RDS instances by modifying the DB engine version, or migrate on-premises SQL Server workloads to take advantage of the new capabilities on managed infrastructure.
How It Works
- Native REST from T-SQL: SQL Server 2025 allows T-SQL code to invoke external REST endpoints directly from the database engine, eliminating the need for middleware layers to call AWS services such as Amazon Bedrock, SageMaker, S3, Lambda, or custom APIs on EC2.
- Native Vector Data Type: A new first-class
vectordata type enables storing and querying vector embeddings directly within SQL Server tables, supporting AI/ML workloads such as semantic search and retrieval-augmented generation (RAG) without external vector stores. - Standard Edition Capacity Expansion: SQL Server 2025 Standard Edition now supports up to 32 cores and 256 GB buffer pool memory (up from 24 cores and 128 GB in prior versions), and gains access to Resource Governor for workload management.
- New Dev-SE Edition: A free Standard Developer Edition (Dev-SE) is available for development and testing scenarios, removing licensing costs in non-production environments.
- Upgrade Path: Existing RDS for SQL Server customers can upgrade to SQL Server 2025 by modifying the DB engine version through the AWS Console, CLI, or API; major version upgrades must be initiated manually after testing.
- Current Minor Version: The latest supported minor version at launch is SQL Server 2025 CU5 (17.0.4045.5), released May 20, 2026.
- Licensing Models: Both License Included (LI) and Bring Your Own Media (BYOM) licensing models are supported for Enterprise, Standard, and Developer editions.
Why It's Important
- AI Integration Without Middleware: The ability to call REST endpoints directly from T-SQL means teams can integrate database workloads with Amazon Bedrock or SageMaker for AI-powered features—such as query advisors or automated performance analysis—without adding application layers or re-architecting existing systems.
- Vector Search in the Database: Native vector type support allows organizations to build semantic search, recommendation engines, and RAG pipelines entirely within SQL Server, reducing architectural complexity and data movement overhead.
- Lower Cost for Standard Edition Workloads: The expanded Standard Edition limits (32 cores, 256 GB memory) and the addition of Resource Governor mean many workloads previously requiring the more expensive Enterprise Edition can now run on Standard, reducing licensing costs.
- Free Development Option: Dev-SE eliminates licensing costs for development and testing environments, lowering the barrier for teams to evaluate SQL Server 2025 features before committing to production deployments.
- On-Premises Migration Incentive: Organizations running SQL Server 2025 on-premises can now migrate to RDS and gain fully managed infrastructure (automated backups, patching, Multi-AZ HA) while retaining access to all new SQL Server 2025 capabilities.
- Event-Driven Database Workflows: Direct REST invocation enables event-driven patterns—such as triggering Lambda functions or posting to S3—from within stored procedures or triggers, opening new integration patterns for legacy database applications.
How It's Different
- REST Calls Without Middleware: Unlike previous SQL Server versions that required CLR integration, linked servers, or external application code to call HTTP endpoints, SQL Server 2025 natively supports REST invocation from T-SQL, simplifying architecture significantly.
- First-Class Vector Support: Prior SQL Server versions required storing vectors as binary blobs or using external vector databases; SQL Server 2025's native
vectortype provides purpose-built storage and query semantics comparable to dedicated vector databases. - Standard Edition Now Rivals Enterprise for Many Workloads: The doubling of Standard Edition memory (128 GB → 256 GB) and core limits, plus the addition of Resource Governor, closes a significant capability gap with Enterprise Edition that existed in all prior versions.
- Free Dev Tier for Standard Features: Previous developer editions mirrored Enterprise Edition features; Dev-SE specifically mirrors Standard Edition, giving developers a cost-free, production-representative environment for Standard Edition workloads.
- Managed Upgrade Path: Unlike on-premises upgrades that require DBA-managed patching and downtime windows, RDS customers can initiate the upgrade via a single API/console modification with RDS handling the underlying infrastructure changes.
When to Prefer It
- AI-Augmented Database Applications: When you want to add LLM-powered features (e.g., natural language query interfaces, automated anomaly explanations) to existing SQL Server applications without introducing new middleware services or changing application code.
- Semantic Search and RAG Pipelines: When building retrieval-augmented generation or similarity search features and you want to keep vector embeddings co-located with relational data in SQL Server rather than managing a separate vector database.
- Cost-Sensitive Workloads Outgrowing Standard Edition Limits: When your workload previously required Enterprise Edition solely due to memory or core count constraints, SQL Server 2025 Standard Edition's expanded limits may now be sufficient at lower cost.
- Event-Driven Workflows from the Database Layer: When you need stored procedures or triggers to invoke AWS Lambda, post to S3, or call SageMaker inference endpoints as part of transactional logic, without adding application-layer orchestration.
- Development and Testing Environments: When you need a production-representative Standard Edition environment for CI/CD pipelines or developer workstations at zero licensing cost, using the new Dev-SE edition.
- On-Premises SQL Server Migration: When running SQL Server on-premises and looking to reduce operational overhead while gaining access to SQL Server 2025 AI features and AWS service integrations on fully managed infrastructure.
- Performance-Sensitive Multi-Tenant Applications: When you need Resource Governor to isolate and prioritize workloads across tenants or applications on Standard Edition, a capability previously requiring an Enterprise Edition upgrade.
Availability
- GA Status: Generally Available as of July 21, 2026, for Enterprise, Standard, and Developer editions on Amazon RDS for SQL Server.
- Supported Minor Version: SQL Server 2025 CU5 (17.0.4045.5) is the current supported version at launch.
- Regional Availability: Specific region availability is not enumerated in the announcement; refer to the [Amazon RDS for SQL Server Pricing page](https://aws.amazon.com/rds/sqlserver/pricing/) for up-to-date regional availability.
- Licensing Models: Both License Included (LI) and Bring Your Own Media (BYOM) via Microsoft License Mobility are supported; Dev-SE is available at no licensing cost for development/testing.
- Pricing Model: Standard RDS for SQL Server on-demand and Reserved Instance pricing applies; Database Savings Plans are eligible for cost optimization.
- Upgrade Path: Major version upgrades from earlier RDS SQL Server versions must be initiated manually; AWS recommends testing upgrades before applying to production instances.
- Limitations: RDS does not provide shell access or direct host access; certain system procedures and advanced privileges remain restricted as part of the managed service model. Multi-AZ deployments use SQL Server DBM or Always On Availability Groups.
