Abstract:The performance of modern database management systems highly relies on hundreds of adjustable knobs. Traditionally, these knobs are manually adjusted by database administrators, a process that is both inefficient and ineffective for tuning large-scale databases in cloud environments. Recent research has explored the use of machine learning techniques to enable the automatic tuning of database configurations. Although most existing learning-based methods achieve satisfactory results on static workloads, they often experience performance degradation and low sampling efficiency in real-world environments. According to our study, this is primarily due to a lack of safety guarantees during the configuration sampling process. To address the aforementioned issues, we propose SafeTune, an online tuning system that adapts to dynamic workloads. Our core idea is to filter out a large number of configurations with potential risks during the configuration sampling process. We employ a two-stage filtering approach: The first stage utilizes a semi-supervised outlier ensemble with feature learning to achieve high-quality feature representation. The second stage employs a ranking-based classifier to refine the filtering process. In addition, to alleviate the cold-start problem, we leverage the historical tuning experience to provide high-quality initial samples during the initialization phase. We conducted comprehensive evaluations on static and dynamic workloads. In comparison to offline baseline methods, SafeTune reduces 95.6%-98.6% unsafe configuration suggestions. In contrast with state-of-the-art methods, SafeTune has improved cumulative performance by 10.5%-46.6% and tuning speed by 15.1%-35.4%.

Simulation on System Dependability Self-Tuning Method Based on Grade Optimization

Research on Grade Optimization Self-tuning Method for System Dependability Based on Autonomic Computing.

System Dependability Self-tuning Algorithm Based on Confidence Field

System Dependability Self-Tuning Model Based on Autonomic Computing

A Self-Optimization Method for System Service Dependability Based on Autonomic Computing

Self-tuning of Software Systems Through Dynamic Quality Tradeoff and Value-Based Feedback Control Loop

Survivability-oriented self-tuning of web systems.

Are your sites down? Requirements-driven self-tuning for the survivability of Web systems

Self-Tuning of Software Systems Through Goal-based Feedback Loop Control

Optimizing System Quality of Service through Rejuvenation for Long-Running Applications with Real-Time Constraints

Reliability analysis and state transfer scheduling optimization of degrading load-sharing system equipped with warm standby components

Towards Online and Safe Configuration Tuning with Semi-supervised Anomaly Detection

Resilience-based design optimization of engineering systems under degradation and different maintenance strategy

Joint optimization of job scheduling, condition-based maintenance planning, and spare parts ordering for degrading production systems

System Optimization Design Under Time Variant Reliability Constraints

Uncertainty Handling In Goal-Driven Self-Optimization - Limiting The Negative Effect On Adaptation

A Real-Time Adaptive Fault Diagnosis Scheme for Dynamic Systems with Performance Degradation

Maintenance Optimization of Multi-Unit Balanced Systems Using Deep Reinforcement Learning

Online Tuning Scheme for Generalized Predictive Controller Via Simulation-Optimization.

Optimization of a multi-node system reliability model

Towards a Degradation-Based Mechanism for Adaptive Overload Control