Databases, warehouses and search stores SPSync copies SharePoint lists into
SPSync connects to many destinations, relational databases, warehouses and lakes, search and vector stores, and the same lists can be replicated to more than one of them, with one replication for each. If yours is not on this list, we are happy to add a custom destination or source for you, book a call and tell us what you need.
For where the data comes from, see sources: SharePoint Online lists and document libraries.
Still deciding whether a replica is the right route at all? SharePoint list to a database compares every way of getting a list into a database, from Excel links and flows to Data Factory and Fabric mirroring.
Destinations at a glance
The licence column names the destination family a licence covers. Embeddings on sync means SPSync writes a vector for each row as it copies. In the last column, listed means the store is on Microsoft's list of Copilot Studio real-time knowledge connectors, a preview feature when we checked on 22 October 2026.
| Destination | Family | Status | Licence | Embeddings on sync | Copilot Studio real-time knowledge |
|---|---|---|---|---|---|
| Google AlloyDB | Relational databases | Shipped | Relational databases | No | Not listed |
| Amazon Aurora | Relational databases | Shipped | Relational databases | No | Not listed |
| Amazon RDS | Relational databases | Shipped | Relational databases | No | Not listed |
| Azure Database for MySQL | Relational databases | Shipped | Relational databases | No | Not listed |
| Azure Database for PostgreSQL | Relational databases | Shipped | Relational databases | No | Not listed |
| Azure SQL Database and Managed Instance | Relational databases | Shipped | Relational databases | No | Listed |
| Dataverse | Relational databases | Shipped | Relational databases | No | Listed |
| Google Cloud SQL | Relational databases | Shipped | Relational databases | No | Not listed |
| Heroku Postgres | Relational databases | Shipped | Relational databases | No | Not listed |
| MariaDB | Relational databases | Shipped | Relational databases | No | Not listed |
| MySQL | Relational databases | Shipped | Relational databases | No | Not listed |
| Neon | Relational databases | Shipped | Relational databases | No | Not listed |
| Oracle Database | Relational databases | Shipped | Relational databases | No | Listed |
| PostgreSQL | Relational databases | Shipped | Relational databases | No | Not listed |
| SQL Server | Relational databases | Shipped | Relational databases | No | Not listed |
| SQLite | Relational databases | Shipped | Included with every product | No | Not listed |
| Supabase | Relational databases | Shipped | Relational databases | No | Not listed |
| Amazon Redshift | Warehouses and lakes | Later, Parquet today | Warehouses and lakes | No | Not listed |
| Amazon S3 | Warehouses and lakes | Shipped | Warehouses and lakes | No | Not listed |
| Azure Blob Storage | Warehouses and lakes | Shipped | Warehouses and lakes | No | Not listed |
| Azure Data Lake Storage Gen2 | Warehouses and lakes | Shipped | Warehouses and lakes | No | Not listed |
| Azure Synapse Analytics | Warehouses and lakes | On the roadmap, Parquet today | Warehouses and lakes | No | Not listed |
| Google BigQuery | Warehouses and lakes | Shipped | Warehouses and lakes | No | Not listed |
| ClickHouse | Warehouses and lakes | Shipped | Warehouses and lakes | No | Not listed |
| Databricks | Warehouses and lakes | Shipped | Warehouses and lakes | No | Listed |
| Delta Lake tables | Warehouses and lakes | Shipped | Warehouses and lakes | No | Not listed |
| DuckDB and MotherDuck | Warehouses and lakes | Shipped | Warehouses and lakes | No | Not listed |
| Apache Iceberg tables | Warehouses and lakes | Shipped | Warehouses and lakes | No | Not listed |
| Microsoft Fabric Lakehouse and OneLake | Warehouses and lakes | On the roadmap, Parquet today | Warehouses and lakes | No | Not listed |
| Parquet files | Warehouses and lakes | Shipped | Warehouses and lakes | No | Not listed |
| Snowflake | Warehouses and lakes | On the roadmap, Parquet today | Warehouses and lakes | No | Listed |
| Azure AI Search | Search and vector stores | Shipped | Search and vector stores | Yes | Listed |
| Elasticsearch | Search and vector stores | Shipped | Search and vector stores | No | Not listed |
| LanceDB | Search and vector stores | Shipped | Search and vector stores | No | Not listed |
| OpenSearch | Search and vector stores | Shipped | Search and vector stores | No | Not listed |
| Postgres with pgvector | Search and vector stores | Shipped | Search and vector stores | Yes | Not listed |
| Qdrant | Search and vector stores | Shipped | Search and vector stores | No | Not listed |
| Weaviate | Search and vector stores | Shipped | Search and vector stores | No | Not listed |
How to choose a destination
Start with whoever reads the data. When a Power BI report, an internal app or an agent with a SQL tool reads it, the usual answer is a relational database your team already runs. That tends to be SQL Server or Azure SQL in a Microsoft shop, and Postgres or MySQL elsewhere. Lookups arrive as key columns and multi-value columns as child tables, so a report joins two lists the way it joins any other tables.
A warehouse or a lake makes sense when the lists feed a data platform you already
have. BigQuery, Databricks, ClickHouse and DuckDB take the tables directly. Parquet
is the route into the platforms SPSync does not write to yet. Each run replaces a
current.parquet file per table, which Snowflake, Synapse, Fabric and
Redshift can load or query today, and each of their pages shows how.
Pick a search or vector store when people, or an agent, look for rows by what the text means. Postgres with pgvector and Azure AI Search are the two where SPSync writes the embedding itself. If your managed Postgres offers the pgvector extension, the pgvector destination can write to it. Elasticsearch, OpenSearch and the other search stores suit keyword and filtered search over the same columns.
For a Copilot Studio agent, the last column of the table matters most. A copy in a store Microsoft lists can be added to the agent as knowledge through that connector. For a copy anywhere else, a custom agent can query it with a read-only SQL tool, the route the AI and agents hub walks through. Either way the agent reads the copy with a database login and needs no Graph permission of its own. Each destination page covers the setup, the type mapping and the limits for that store.
Relational databases
- SharePoint to Google AlloyDB
- SharePoint to Amazon Aurora (Postgres, MySQL)
- SharePoint to Amazon RDS (SQL Server, Postgres, MySQL engines)
- SharePoint to Azure Database for MySQL
- SharePoint to Azure Database for PostgreSQL
- SharePoint to Azure SQL Database and Managed Instance
- SharePoint to Dataverse (for Power Apps that need delegation)
- SharePoint to Google Cloud SQL (SQL Server, Postgres, MySQL)
- SharePoint to Heroku Postgres
- SharePoint to MariaDB
- SharePoint to MySQL
- SharePoint to Neon
- SharePoint to Oracle Database
- SharePoint to PostgreSQL
- SharePoint to SQL Server
- SharePoint to SQLite (local file, desktop analytics)
- SharePoint to Supabase
Warehouses and lakes
- SharePoint to Amazon Redshift
- SharePoint to Amazon S3 (Parquet)
- SharePoint to Azure Blob Storage (Parquet)
- SharePoint to Azure Data Lake Storage Gen2
- SharePoint list to Azure Synapse Analytics
- SharePoint to Google BigQuery
- SharePoint to ClickHouse
- SharePoint to Databricks (Delta, Unity Catalog)
- SharePoint to Delta Lake tables on blob
- SharePoint to DuckDB and MotherDuck
- SharePoint to Apache Iceberg tables
- SharePoint list to Microsoft Fabric Lakehouse and OneLake
- SharePoint to Parquet files on Azure Blob, S3 or local disk
- SharePoint to Snowflake
Search and vector stores
Postgres with pgvector and Azure AI Search receive the replicated columns and an embedding per row that SPSync generates on each sync, from OpenAI, Azure OpenAI or a local model. Elasticsearch, OpenSearch, LanceDB, Qdrant and Weaviate receive the replicated columns and no vector; they fit keyword and filtered search, and a vector you add belongs to your own pipeline.