AWS Introduces Multi-Dataset Relationships in QuickSight
Amazon Web Services has launched Multi-Dataset Relationships for Amazon QuickSight, a feature that enables users to define logical associations between separate datasets and perform runtime joins at query time — eliminating the need to pre-flatten tables. Announced on the AWS Machine Learning blog, this capability allows businesses to keep each data source as an independent QuickSight dataset while declaring relationships inside a QuickSight Topic.
This shift from ETL-heavy preprocessing to on-the-fly schema joining marks a significant architectural change for business intelligence (BI) workflows, particularly for organizations that manage complex, multi-source data environments.
What Happened and How It Works
Previously, QuickSight users had to flatten related tables (e.g., merging sales transactions with customer profiles into a single dataset) before visualization. This approach required additional data pipeline maintenance, storage overhead, and increased the risk of stale or duplicated data. The new Multi-Dataset Relationships feature works by letting analysts define relationships — such as one-to-many or many-to-one — between datasets within a QuickSight Topic. When a user builds a visual or runs a query, QuickSight performs a runtime join across the related datasets, pulling only the necessary columns and rows at that moment.
According to AWS, this capability is powered by SPICE (Super-fast, Parallel, In-memory Calculation Engine), meaning the runtime joins occur in memory without writing intermediate tables to disk. The feature supports common join types including inner, left, and full outer joins. Early adopters, including e-commerce and fintech firms, report reducing data preparation time by up to 50% for BI projects with five or more source tables.
Why It Matters for Developers and Data Engineers
For AI developers and data engineers, Multi-Dataset Relationships eliminates a recurring pain point: building and maintaining data marts specifically for QuickSight consumption. Instead of constructing flattened views in Redshift, Athena, or S3, teams can now model data more naturally, preserving referential integrity and reducing duplication. This is particularly relevant for machine learning pipelines where training data often requires joining multiple tables — now, those same raw tables can be used directly for BI without additional preprocessing.
From a performance perspective, AWS claims that runtime joins in SPICE are optimized for sub-second response times on datasets under 100 million rows, though performance degrades with very large fact tables or complex multi-hop joins. Developers should consider cardinality: high-cardinality columns in join keys may impact memory usage. The feature currently supports up to 10 related datasets per Topic.
Implications for Business Intelligence Workflows
Business analysts and non-technical users benefit from a more intuitive data modeling experience. Instead of needing SQL skills or a data engineer to create flat files, analysts can simply check which datasets relate to each other and start building dashboards. AWS has also introduced a best practices guide detailing how to avoid circular relationships and when to use star schemas versus snowflake schemas within Topics.
For enterprises with strict governance requirements, Multi-Dataset Relationships inherits QuickSight’s row-level security (RLS) and column-level permissions, ensuring that runtime joins respect existing access controls. This is critical for regulated industries like healthcare and finance, where data siloes are often intentional for compliance reasons.
What This Means for the AI and Cloud Ecosystem
This launch aligns with a broader industry trend: reducing ETL complexity while maintaining query performance. Competitors like Tableau and Looker have supported multi-table joins for years, but QuickSight’s move is significant because it integrates deeply with AWS’s data lake and analytics ecosystem — Glue, Lake Formation, and Athena. AI developers building recommendation systems or anomaly detection models can now use QuickSight Topics as a semantic layer, making it easier to expose BI insights alongside ML model outputs.
That said, there are limitations. Runtime joins do not support aggregations across related datasets without explicit measures, and users cannot write custom SQL within Topics. For advanced use cases, AWS recommends using Athena Federated Query or Redshift Spectrum. Additionally, the feature is currently available only in AWS regions supporting QuickSight Standard and Enterprise editions, with pricing included in existing SPICE capacity.
Developer Takeaway: Best Practices
AWS recommends several best practices for developers adopting Multi-Dataset Relationships. First, ensure join keys are indexed in the underlying data sources to maximize SPICE lookup speed. Second, avoid many-to-many relationships, which can cause unexpected row multiplication. Third, use Topics as a semantic layer for business terms rather than technical column names — for example, map a column named 'cust_id' to 'Customer Identifier'. Finally, test performance with representative data volumes before rolling out to production dashboards.
As of May 2026, Multi-Dataset Relationships is generally available. Developers can enable it via the QuickSight console or API. AWS has also published a sample GitHub repository with Python scripts for automating Topic creation and relationship definitions.
Conclusion
AWS QuickSight’s Multi-Dataset Relationships removes a major barrier to agile BI: the mandatory flattening of datasets. By enabling runtime joins, the service reduces pipeline complexity and accelerates time-to-insight for AI-driven organizations. While not a replacement for heavy-duty ETL tools, it offers a pragmatic middle ground for teams that want to keep their data models clean while enabling fast, ad-hoc analysis.
Source: AWS Machine Learning. This article was produced with AI assistance and reviewed for accuracy. Editorial standards.
