IMLane: Composable Framework for Efficient AI Function Execution in Database Engine
| Paper | IMLane: Composable Framework for Efficient AI Function Execution in Database Engine (PDF) |
| Implementation | Code |
| Venue | VLDB 2026 |
Abstract
Efficient execution of artificial intelligence (AI) functions within the database engine has become an essential requirement for AI-driven data analysis workflows. During the development of supporting AI functions in OceanBase, we identified two performance bottlenecks incurred by the internal design of the database engine. First, the Python UDF-based AI function is executed via thread-level parallelism, which causes ineffective parallel execution due to limitations of CPython runtime. Second, the scheduling of the AI function is tightly coupled with the database engine’s scheduler. This scheduler fails to adapt to the diverse compute resource requirements of AI functions in different user scenarios, leading to the under-utilization of available compute resources. To address these common bottlenecks, we present a composable framework, called IMLane, that equips current database engines with efficient AI function execution. IMLane incorporates process-level AI function execution for effective parallel execution with low data transfer overhead, and decoupled AI function scheduling to match compute resource requirements. Our experiments on OceanBase and DuckDB show 7.48x and 5.04x improvements on average, respectively, after integrating IMLane. Moreover, IMLane demonstrates performance superiority over alternative solutions.
