Are you using DatabaseSessionService for session management or defining your own tables?

[ecanlar]

Hi!
I'm building a production assistant using ADK with PostgreSQL session persistence and I'd like to know how others are approaching session and user management.

Specifically:

• Are you using the built-in DatabaseSessionService to handle sessions, or have you defined your own database tables and custom session logic?
• If you use DatabaseSessionService, do you extend it or wrap it for user-level session isolation?

Currently using DatabaseSessionService with a per-user session_id passed via WebSocket, but wondering if there's a more idiomatic pattern the community has converged on.

Thanks!

[adk-bot]

Response from ADK Answering Agent (experimental, answer may be inaccurate)

TLDR: Your current approach using DatabaseSessionService is the correct and idiomatic way to handle session persistence in ADK. The service is designed for user-level isolation using the user_id, so you don't need to extend or wrap it for this purpose.

---

Hi! That's a great question. It's good to see you're thinking about production-level concerns like session management.

You can be confident in your current approach. Using DatabaseSessionService is indeed the standard and recommended pattern for handling persistent sessions when using a self-managed database like PostgreSQL with ADK.

To answer your specific questions:

1. DatabaseSessionService vs. Custom Logic: The community and documentation strongly recommend using the built-in session services. DatabaseSessionService is the go-to for production environments with relational databases. The main alternatives are InMemorySessionService (intended for development and testing) and VertexAiSessionService (for GCP-managed persistence). Building a custom session management system from scratch is generally not necessary, as the built-in services are robust and well-integrated.

2. User-Level Session Isolation: You do not need to extend or wrap DatabaseSessionService to achieve user-level isolation. This is a core feature of the service. ADK's session management uses a three-level hierarchy to uniquely identify a session: app_name, user_id, and session_id. [1] This structure is specifically designed for multi-tenancy and ensures that each user's data is kept separate.

   When DatabaseSessionService creates its tables, the sessions table uses a composite primary key of (app_name, user_id, id). This enforces session isolation at the database level. [2] Your application logic simply needs to provide the correct user_id when creating or retrieving a session, and the service handles the rest.

Your method of passing a session_id via WebSocket is a common pattern for the client to specify which conversation it wants to continue. On the backend, you would use this session_id along with the authenticated user's user_id to fetch the correct session from the DatabaseSessionService.

Hopefully, this confirms you're on the right track!

---

References
[1] ADK Streaming Developer Guide - https://google.github.io/adk-docs/streaming/dev-guide/part1/
[2] PostgreSQL Session Service Sample - https://github.com/google/adk-python/blob/main/contributing/samples/postgres_session_service/README.md