OpenSearch

OpenSearch is an open-source search and analytics engine with a k-NN (k-Nearest Neighbor) plugin for approximate vector search. The synaptic-rag crate (with feature opensearch) implements the VectorStore trait using OpenSearch's HNSW-based k-NN indexing.

Setup

Add the feature flag to your Cargo.toml:

[dependencies]
synaptic = { version = "0.4", features = ["opensearch"] }

Run OpenSearch locally with Docker:

docker run -d --name opensearch \
  -p 9200:9200 -p 9600:9600 \
  -e "discovery.type=single-node" \
  -e "plugins.security.disabled=true" \
  opensearchproject/opensearch:latest

Usage

use synaptic::opensearch::{OpenSearchConfig, OpenSearchVectorStore};
use synaptic::core::VectorStore;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let config = OpenSearchConfig::new("http://localhost:9200", "my_index", 1536)
        .with_credentials("admin", "admin");
    let store = OpenSearchVectorStore::new(config);

    // Create the index with k-NN mappings (idempotent)
    store.initialize().await?;

    // Add documents
    // store.add_documents(docs, &embeddings).await?;

    // Search
    // let results = store.similarity_search("query text", 5, &embeddings).await?;

    Ok(())
}

Amazon OpenSearch Service

For Amazon OpenSearch Service, set the endpoint to your AWS-provisioned domain:

let config = OpenSearchConfig::new(
    "https://my-domain.us-east-1.es.amazonaws.com",
    "my_index",
    1536,
);

Configuration