SAM: A Stability-Aware Cache Manager for Multi-Tenant Embedded Databases

2025-07-30

AI Summary

Multi-tenant database co-location on edge nodes induces severe cache contention, while conventional cache management schemes exhibit poor stability and weak adaptability under dynamic workloads. This paper proposes a self-managing cache framework centered on stability, introducing a novel dual-factor model: the H-factor (Historical Stability Efficiency) and the V-factor (Future Marginal Gain), balanced via an adaptive weighting mechanism to jointly optimize exploitation and exploration. Based on this model, we design AURA—a dynamic cache scheduling algorithm that integrates historical performance prediction with real-time workload analysis to enable scalable, low-latency decision-making. Experimental evaluation on a 120-node cluster demonstrates near-constant decision latency, significantly higher throughput than state-of-the-art baselines, and strong robustness against both sudden load spikes and cache pollution attacks. Moreover, service quality remains highly predictable across diverse operational scenarios.

Technology Category

• Planning, Routing, and Scheduling: Scheduling under Uncertainty
• Machine Learning: Hardware-aware ML
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Security and Privacy: Large-scale security measurements

Abstract

The co-location of multiple database instances on resource constrained edge nodes creates significant cache contention, where traditional schemes are inefficient and unstable under dynamic workloads. To address this, we present SAM, an autonomic cache manager powered by our novel AURA algorithm. AURA makes stability a first-class design principle by resolving the exploitation-exploration dilemma: it achieves this by synthesizing two orthogonal factors, which we introduce as: the H-factor, representing a database's proven, historically stable efficiency (exploitation), and the V-factor, representing its empirically estimated marginal gain for future improvements (exploration). This dual-factor model, governed by an adaptive weight, enables SAM to achieve sustained high performance through strategic stability and robustness in volatile conditions. Extensive experiments against 14 diverse baselines demonstrate SAM's superiority. It achieves top-tier throughput while being uniquely resilient to complex workload shifts and cache pollution attacks. Furthermore, its decision latency is highly scalable, remaining nearly constant as the system grows to 120 databases. Crucially, SAM achieves superior decision stability -- maintaining consistent optimization directions despite noise, avoiding performance oscillations while ensuring predictable Quality of Service. These results prove that a principled, stability-aware design is essential for sustained high performance in real-world, large-scale systems.

Problem

• Addresses cache contention in multi-tenant embedded databases
• Resolves exploitation-exploration dilemma in dynamic workloads
• Ensures stability and performance under volatile conditions

Innovation

• AURA algorithm for stability-aware cache management
• Dual-factor model (H-factor and V-factor) for exploitation-exploration balance
• Scalable decision latency with constant performance growth