Milvus Destination for Fivetran

Category: Data Integration & Migration
Brand: Fivetran
Type: Destination (Vector Database)

Milvus Destination for Fivetran lets you use Fivetran ELT pipelines to load data into Milvus or Zilliz Cloud as a vector database, supporting AI and similarity search workloads such as semantic search, RAG, and multi‑modal search.

Overview

• Integrates Fivetran with Milvus, an open‑source vector database built for GenAI applications.
• Supports both Milvus and Zilliz Cloud (fully managed Milvus) as destinations.
• Designed to power AI use cases including semantic search, Retrieval‑Augmented Generation (RAG), and multi‑modal search by syncing data into a vector database.
• Best‑practice recommendation is to run Milvus in Zilliz Cloud for quickest setup.

Features

Destination capabilities

• Milvus and Zilliz Cloud support
  Fivetran can write to self‑managed Milvus or Zillan Cloud clusters as the destination.

• GenAI and vector search workloads
  Data is loaded into Milvus as vectors suitable for high‑performance similarity search and AI applications (semantic search, RAG, multi‑modal use cases).

Source connector support

Only the following Fivetran connectors are supported as sources for this destination:

• Snowflake (database connector)
• BigQuery (database connector)
• Databricks (database connector)

The documentation also references guidance on how to sync data from sources not directly supported by the Milvus destination, implying workflows or intermediate steps for unsupported sources.

Deployment model

• SaaS deployment supported
  Milvus destination is supported when using Fivetran’s SaaS Deployment model.

Setup and configuration

• Step‑by‑step setup guide
  A dedicated Milvus setup guide walks through connecting a Milvus vector database (or Zilliz Cloud cluster) as a Fivetran destination.

• Partner‑built destination
  The Milvus destination is identified as partner‑built, with documentation and implementation maintained by the Milvus/Zilliz team.

Data type mapping

Fivetran’s standard data types are mapped to Milvus data types as follows:

| Fivetran Data Type | Milvus Destination Data Type | | -------------------| -----------------------------| | BOOLEAN | BOOLEAN | | SHORT | INT16 | | INT | INT32 | | LONG | INT64 | | FLOAT | FLOAT | | DOUBLE | DOUBLE | | DECIMAL | VARCHAR | | LOCALDATE | VARCHAR |

These mappings define how source schema fields are stored in Milvus when synced via Fivetran.

Integrations

• Vector Database: Milvus, Zilliz Cloud
• Supported Source Databases:
  • Snowflake
  • BigQuery
  • Databricks

Pricing

• The provided documentation content does not specify any dedicated pricing details or plans specific to the Milvus destination.
• Fivetran generally operates on usage‑based pricing, described separately in Fivetran’s “Usage‑Based Pricing” documentation, but no Milvus‑specific pricing tiers or constraints are mentioned in the extracted content.