Skip to main content
Loading

Quickstart for Vector

This page describes how to get up and running quickly with Aerospike Vector Search (AVS) using Docker.

tip

Request a 60-day free trial feature-key file with the vector-service option enabled to install AVS locally.

  1. Clone the examples repository.

    The AVS repository contains several pre-configured example applications for deploying and using AVS.

    git clone https://github.com/aerospike/aerospike-vector
  1. Copy your feature-key file.

    Your feature-key file must be named features.conf and be located in following location.

    cp ~/Downloads/features.conf ./aerospike-vector/prism-image-search/container-volumes/avs/etc/aerospike-vector-search
  2. Index your photos.

    Use the prism-image-search application to index your photos. This application uses the OpenAI CLIP model to create embeddings for your photos. These embeddings are indexed in AVS, and you can perform a semantic search on the index.

    Use the following command to copy .jpeg and .jpg files from your ~/Pictures directory to be indexed.

    rsync -av --include='*/' --include='*.jp*' --exclude='*' ~/Pictures ./aerospike-vector/prism-image-search/container-volumes/prism/images/static/data
  3. Start the database, AVS, and the search application.

    docker-compose -f ./aerospike-vector/prism-image-search/docker-compose.yml up -d
  4. After a few minutes, navigate to http://localhost:8080/search and search your photos.

Next Steps

After completing the quickstart, you can do the following:

  • Install asvec: The asvec tool is a command line interface (CLI) for managing AVS.

  • Create a data pipeline: This page outlines various approaches for generating vector embeddings that you can use on your own data.