Skip to main content

PgVector Store

Banner Spring AI Icon

🚀 Fun with PgVectorStore, OpenAI, and Ollama in Spring AI​

So, you’ve decided to dive into the magical world of Spring AI, Postgres, and PgVectorStore? Buckle up! We’re about to turn your database into a vector-storing, AI-boosted, RAG-ready beast! 🦾🔥

🏗️ Step 1: Installing the Postgres PgVector Database​

PgVector is like that cool kid in the database world who can store vectors and find similar ones in a snap! Let’s get it up and running.

🐳 1.1. Using Docker​

Why complicate things when Docker makes life easy? Run this:

docker run -it --rm --name postgres -p 5432:5432 \
  -e POSTGRES_USER=postgres \
  -e POSTGRES_PASSWORD=postgres \
  -e VECTOR_STORE_TYPE=pgVector \
  pgvector/pgvector:pg16

Or, if you’re fancy and use Docker Compose, here’s your setup:

services:
  postgres:
    image: pgvector/pgvector:pg16
    container_name: postgres
    ports:
      - "5432:5432"
    environment:
      POSTGRES_USER: postgres
      POSTGRES_PASSWORD: postgres
      VECTOR_STORE_TYPE: pgVector

🚨 Heads up! If you already have Postgres running, don’t forget to enable the extensions vector, hstore, and uuid-ossp.

⚙️ Step 2: Configuring PgVector with Spring AI​

Spring AI does the heavy lifting for us! Just set:

spring.ai.vectorstore.pgvector.initialize-schema=true

And it will magically run this schema setup:

CREATE EXTENSION IF NOT EXISTS vector;
CREATE EXTENSION IF NOT EXISTS hstore;
CREATE EXTENSION IF NOT EXISTS "uuid-ossp";

CREATE TABLE IF NOT EXISTS vector_store (
  id uuid DEFAULT uuid_generate_v4() PRIMARY KEY,
  content text,
  metadata json,
  embedding vector(1536)
);

CREATE INDEX ON vector_store USING HNSW (embedding vector_cosine_ops);

💡 Pro Tip: You can customize the index type, distance type, and dimensions via application.properties!

🤖 Step 3: Configuring PgVectorStore with OpenAI Embedding Models​

Want to make your app super smart with OpenAI? Let's do this! 🧠

3.1. Add Dependencies​

<dependency>
  <groupId>org.springframework.ai</groupId>
  <artifactId>spring-ai-openai-spring-boot-starter</artifactId>
</dependency>
<dependency>
  <groupId>org.springframework.ai</groupId>
  <artifactId>spring-ai-pgvector-store-spring-boot-starter</artifactId>
</dependency>

3.2. Properties Configuration​

spring.ai.openai.api-key=${OPENAI_API_KEY}
spring.ai.openai.chat.model=gpt-4
spring.ai.openai.embedding.model=text-embedding-ada-002

3.3. Demo App - Let's Test It​

@SpringBootApplication
public class App implements CommandLineRunner {

  public static void main(String[] args) {
    SpringApplication.run(App.class);
  }

  private final VectorStore vectorStore;

  public App(VectorStore vectorStore) {
    this.vectorStore = vectorStore;
  }

  @Override
  public void run(String... args) {
    List<Document> documents = List.of(
      new Document("Java is a high-level programming language."),
      new Document("It is used for enterprise apps and Android development."),
      new Document("Java has strong typing and automatic memory management.")
    );
    
    vectorStore.add(documents);
    
    List<Document> results = vectorStore
      .similaritySearch(SearchRequest.query("programming language").withTopK(1));
    
    results.forEach(doc -> System.out.println(doc.getContent()));
  }
}

🎉 Output:

Java is a high-level programming language.

🤯 Step 4: Configuring PgVectorStore with Ollama Embedding Models​

Wanna run LLMs locally? Meet Ollama! 🏡🦙

4.1. Run Ollama with Docker​

docker run -d --gpus all -v ollama:/root/.ollama -p 11434:11434 --name ollama ollama/ollama

Now download & run a model:

ollama run mistral

4.2. Add Dependencies​

<dependency>
  <groupId>org.springframework.ai</groupId>
  <artifactId>spring-ai-ollama-spring-boot-starter</artifactId>
</dependency>

4.3. Properties Configuration​

spring.ai.ollama.embedding.options.model=mistral
spring.ai.ollama.chat.options.model=mistral

4.4. Demo App - Ollama in Action​

Same as OpenAI, but powered by Mistral! ⚡

🚨 Step 5: "Oops, Too Many Dimensions!"​

Error: column cannot have more than 2000 dimensions

Why? PgVector can’t handle >2000 dimensions for certain index types.

Fix Options​

  1. Use a different vector store (like ChromDB).
  2. Set index-type=none (⚠️ Only for dev, NOT for production!).
  3. Reduce dimensions under 2000.

Example:

spring.ai.vectorstore.pgvector.index-type=none # Dev only!
spring.ai.vectorstore.pgvector.dimensions=4096

🎯 Summary - What We Achieved Today​

  • ✅ Installed Postgres with PgVector extension.
  • ✅ Configured Spring AI for automatic setup.
  • ✅ Connected PgVector with OpenAI & Ollama.
  • ✅ Built a cool Spring Boot app to store & retrieve embeddings!
  • ✅ Solved the 2000-dimension error like a boss. 😎

🚀 Happy Learning & Happy Coding! 🎉