GitHub user wangzhigang1999 added a comment to the discussion: [KPIP] Data
Agent Engine — AI-Powered Autonomous Data Analysis for Kyuubi
I suggests sticking with the former approach—having the tool connect back to
the Kyuubi Server via JDBC/REST, rather than connecting directly to the engine.
By routing through the Kyuubi Server, the Data Agent can fully leverage
existing capabilities like AuthZ (Ranger integration), query auditing,
multi-tenant pooling, and engine lifecycle management. The key benefit is that
this requires no changes on the Server side.
If we connect directly to the engine, it would bypass the Server's
authorization and session management. Then, the agent would need to handle
engine discovery and lifecycle on its own, which might break the current
isolation model.
Here is the proposed flow:
```text
Client ──JDBC──▶ Kyuubi Server ──▶ Data Agent Engine
│
ReAct Loop
LLM ←→ Tools
│
sql_query tool
│
JDBC Driver
│
Kyuubi Server ──▶ Spark/Trino/Hive Engine
```
What do you think about this direction?
GitHub link:
https://github.com/apache/kyuubi/discussions/7373#discussioncomment-16389859
----
This is an automatically sent email for [email protected].
To unsubscribe, please send an email to:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
