ClickHouse Vector Database

Use ClickHouse as a vector database for your Knowledge Base.

Setup

uv pip install -U "clickhouse-connect[async]" pypdf openai sqlalchemy agno

Run ClickHouse with Docker:

docker run -d \
  -e CLICKHOUSE_DB=ai \
  -e CLICKHOUSE_USER=ai \
  -e CLICKHOUSE_PASSWORD=ai \
  -e CLICKHOUSE_DEFAULT_ACCESS_MANAGEMENT=1 \
  -v clickhouse_data:/var/lib/clickhouse/ \
  -v clickhouse_log:/var/log/clickhouse-server/ \
  -p 8123:8123 \
  -p 9000:9000 \
  --ulimit nofile=262144:262144 \
  --name clickhouse-server \
  clickhouse/clickhouse-server

Set your OpenAI API key. The examples below use the default OpenAI embedder and model.

export OPENAI_API_KEY=xxx

Example

agent_with_knowledge.py
from agno.agent import Agent
from agno.knowledge.knowledge import Knowledge
from agno.db.sqlite import SqliteDb
from agno.vectordb.clickhouse import Clickhouse

knowledge=Knowledge(
    vector_db=Clickhouse(
        table_name="recipe_documents",
        host="localhost",
        port=8123,
        username="ai",
        password="ai",
    ),
)

knowledge.insert(
  url="https://agno-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf"
)

agent = Agent(
    db=SqliteDb(db_file="agno.db"),
    knowledge=knowledge,
    # Enable the agent to search the knowledge base
    search_knowledge=True,
    # Enable the agent to read the chat history
    read_chat_history=True,
)

agent.print_response("How do I make pad thai?", markdown=True)
agent.print_response("What was my last question?", stream=True)

Async Support ⚡

ClickHouse also supports asynchronous operations, enabling concurrency and leading to better performance.

async_clickhouse.py
import asyncio

from agno.agent import Agent
from agno.knowledge.knowledge import Knowledge
from agno.db.sqlite import SqliteDb
from agno.vectordb.clickhouse import Clickhouse

agent = Agent(
    db=SqliteDb(db_file="agno.db"),
    knowledge=Knowledge(
        vector_db=Clickhouse(
            table_name="recipe_documents",
            host="localhost",
            port=8123,
            username="ai",
            password="ai",
        ),
    ),
    # Enable the agent to search the knowledge base
    search_knowledge=True,
    # Enable the agent to read the chat history
    read_chat_history=True,
)

async def main():
    # Keep ingestion and querying on the same event loop
    await agent.knowledge.ainsert(
        url="https://agno-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf"
    )
    await agent.aprint_response("How to make Tom Kha Gai", markdown=True)

if __name__ == "__main__":
    asyncio.run(main())

Use ainsert() and aprint_response() methods with asyncio.run() for non-blocking operations in high-throughput applications.

ClickHouse Params

ParameterTypeDefaultDescription
table_namestr-Name of the table to store vectors and metadata in Clickhouse
hoststr-Hostname of the Clickhouse server
usernameOptional[str]NoneUsername for Clickhouse authentication
passwordstr""Password for Clickhouse authentication
portint0Port number for Clickhouse connection
database_namestr"ai"Name of the database to use in Clickhouse
dsnOptional[str]NoneDSN string for Clickhouse connection
compressstr"lz4"Compression algorithm to use
clientOptional[Client]NoneOptional pre-configured Clickhouse client
asyncclientOptional[AsyncClient]NoneOptional pre-configured async Clickhouse client
embedderOptional[Embedder]OpenAIEmbedder()Embedder instance to generate embeddings
distanceDistanceDistance.cosineDistance metric to use for similarity search
indexOptional[HNSW]HNSW()HNSW index configuration for vector similarity search