Milvus, the open source cloud native vector database for AI and similarity search, preinstalled as the official standalone stack with the etcd metadata store, the MinIO object backend and the Milvus server running as a system service. Authentication is enabled and the root password is rotated on first boot. Backed by 24/7 cloudimg support.
This is a repackaged open source software product wherein additional charges apply for cloudimg support services.
Overview
Milvus is a high performance, cloud native open source vector database built for similarity search and AI applications. It stores, indexes and searches billions of embedding vectors with low latency, powering retrieval augmented generation, semantic search, recommendation, image and video search, anomaly detection and other AI workloads. This image delivers the official Milvus standalone stack fully installed and configured as a system service, so a production grade vector database is running within minutes of launch.
Application Stack
The appliance is the official Milvus standalone Docker Compose stack: an etcd service for metadata and coordination, a MinIO service for the object storage backend and the Milvus standalone server itself, which exposes the gRPC and HTTP APIs. Docker Engine and the Docker Compose plugin are installed from the official Docker package repository, and a systemd service starts the stack on boot and restarts it on failure.
Vector Database
Connect your application with the Milvus SDK (pymilvus and the Go, Java and Node clients) on the gRPC port to create collections, insert embedding vectors with scalar metadata, build vector indexes such as IVF, HNSW and DiskANN, and run approximate nearest neighbour search with metadata filtering. Milvus supports dense and sparse vectors, hybrid search, scalar filtering and a rich set of index and metric types, so you can tune for recall, latency and cost.
Dedicated Storage Tier
The etcd metadata store, the MinIO object data and the Milvus segment data all live on a separate, independently resizable data volume, so the stateful tier can be grown without disturbing the rest of the instance.
Secure First Boot
Authentication is enabled out of the box. On the first boot of your instance a one shot service rotates the database root password, unique to that instance, and rotates the object store secret, then writes the root password to a file only the root user can read. No shared or default credentials ship in the image.
Ready To Use
Connect over SSH and read the welcome notes for the rotated root credentials and the connection endpoints. Point your Milvus client at the instance on the gRPC port with the root user and the generated password, then create your first collection and start inserting and searching vectors. The HTTP port also serves the built in WebUI for inspecting collections and database health.
cloudimg Support
24/7 technical support by email and chat. Help with deployment, collection and index design, embedding ingestion, search tuning, metadata filtering, scaling, backups and client integration.
Use Cases
The vector store behind a retrieval augmented generation pipeline. Semantic and similarity search over text, image, audio or video embeddings. A recommendation and personalisation backend. Duplicate detection and anomaly detection. A self hosted open source alternative to hosted vector database services.
Milvus is a trademark of LF Projects, LLC. All other product and company names are trademarks or registered trademarks of their respective holders. Use of them does not imply any affiliation with or endorsement by them.