Piece of CAKE: Adaptive Execution Engines via Microsecond-Scale Learning

TL;DR

CAKE is an adaptive system that uses microsecond-scale learning with counterfactuals to select optimal database kernels per data morsel, significantly reducing workload latency.

Contribution

It introduces CAKE, a novel adaptive kernel selection system leveraging counterfactual multi-armed bandits and regret trees for low-latency decision making.

Findings

Up to 2x reduction in workload latency

Effective kernel selection per data morsel

Low-latency decision-making with regret trees

Abstract

Low-level database operators often admit multiple physical implementations ("kernels") that are semantically equivalent but have vastly different performance characteristics depending on the input data distribution. Existing database systems typically rely on static heuristics or worst-case optimal defaults to select these kernels, often missing significant performance opportunities. In this work, we propose CAKE (Counterfactual Adaptive Kernel Execution), a system that learns to select the optimal kernel for each data "morsel" using a microsecond-scale contextual multi-armed bandit. CAKE circumvents the high latency of traditional reinforcement learning by exploiting the cheapness of counterfactuals -- selectively running multiple kernels to obtain full feedback -- and compiling policies into low-latency regret trees. Experimentally, we show that CAKE can reduce end-to-end workload latency by up to 2x compared to state-of-the-art static heuristics.