Control Theory: a Foundational Technique for Self Managing Databases

Lightstone, Sam; Surendra, Maheswaran; Diao, Yixin; Parekh, Sujay; Hellerstein, Joseph L.; Rose, Kevin C.; Storm, Adam; Garcia-Arellano, Christian

doi:10.1109/icdew.2007.4401022

Cited by 16 publications

(10 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Thus although there is some work on the use of control theory (e.g. [33]) or economic approaches (e.g. [47,48]) for adapting query processing behavior or resource usage in databases at runtime, the principal focus here is on action strategies.…”

Section: Related Workmentioning

confidence: 99%

Utility-driven adaptive query workload execution

Paton¹,

Aragão²,

Fernandes³

2012

Future Generation Computer Systems

View full text Add to dashboard Cite

Workload management coordinates access to and use of shared computational resources; adaptive workload execution revises resource allocation decisions dynamically in response to feedback about the progress of the workload or the behavior of the resources. Where the workload contains or consists of database queries, adaptive query processing (AQP) changes the way in which a query is being evaluated while the query is running. In parallel environments, available adaptations may change the allocation of query fragments to a machine, for example to remove load imbalance or change the parallelism level. Most AQP strategies act on individual queries with the objective of reducing response times. However, where adaptations affect the usage of shared resources, or the principal goal is to meet quality of service targets rather than to minimize overall response times, locally beneficial decisions may have globally detrimental effects. This paper describes the use of utility functions to coordinate adaptations that assign resources to query fragments from multiple queries, and demonstrates how a common framework can be used to support different objectives, specifically to minimize overall query response times and to maximize the number of queries meeting quality of service goals. Experiments using simulation compare the use of utility functions with the more common heuristic control strategies, demonstrating situations in which significant benefits can be obtained.

show abstract

Section: Related Workmentioning

confidence: 99%

Utility-driven adaptive query workload execution

Paton¹,

Aragão²,

Fernandes³

2012

Future Generation Computer Systems

View full text Add to dashboard Cite

show abstract

“…[3] controls QoS of services by resource entitlement adaptation in resource-partitioned and virtualized environments. Recent versions of the IBM DB2 database apply control theory to memory tuning and database utility throttling [5]. [16] uses control models to solve hot spot contention in network-level congestion control.…”

Section: Related Workmentioning

confidence: 99%

Qualitative performance control in supervised IT infrastructures

Paljak

Egel

Toth

et al. 2010

2010 International Conference on Dependable Systems and Networks Workshops (DSN-W)

View full text Add to dashboard Cite

Performability control of IT systems still lacks theoretically well-founded approaches that fit well to enterprise system management solutions. We propose a methodology for designing compact qualitative, statebased predictive performability control that use instrumentation provided by typical system monitoring frameworks. We identify the main systemic insufficiencies of current monitoring tools that hinder designing trustworthy fine-granular controls.

show abstract

“…Since repartitioning is not expensive for PLP (except for PLP-Partition), it can repartition again very quickly to alter the bad effects of a previous bad partitioning choice. Rather than directly aiming to reach the target load, a more robust technique would be to employ control theory while converging to the target load [31]. Control theory can increase the robustness of our algorithm, prevent the system from repartitioning unnecessarily and/or resulting with wrong partitions, and reduce the downtime faced by PLP-Partition during repartitioning.…”

Section: Using Control Theory For Load Balancingmentioning

confidence: 99%

Scalable and dynamically balanced shared-everything OLTP with physiological partitioning

et al. 2012

View full text Add to dashboard Cite

Scaling the performance of shared-everything transaction processing systems to highly parallel multicore hardware remains a challenge for database system designers. Recent proposals alleviate locking and logging bottlenecks in the system, leaving page latching as the next potential problem. To tackle the page latching problem, we propose physiological partitioning (PLP). PLP applies logical-only partitioning, maintaining the desired properties of sharedeverything designs, and introduces a multi-rooted B+Tree index structure (MRBTree) that enables the partitioning of the accesses at the physical page level. Logical partitioning and MRBTrees together ensure that all accesses to a given index page come from a single thread and, hence, can be entirely latch free; an extended design makes heap page accesses thread private as well. Moreover, MRBTrees offer an infrastructure for easy repartitioning and allow us to have a lightweight dynamic load balancing mechanism (DLB) on top of PLP. Profiling a PLP prototype running on different multicore machines shows that it acquires 85 and 68 % fewer contentious critical sections, respectively, than an optimized conventional design and one based on logical-only partitioning. PLP also improves performance up to almost 50 % over the existing systems, while DLB enhances the system with rapid and robust behavior in both detecting and handling load imbalances.

show abstract

Control Theory: a Foundational Technique for Self Managing Databases

Cited by 16 publications

References 10 publications

Utility-driven adaptive query workload execution

Utility-driven adaptive query workload execution

Qualitative performance control in supervised IT infrastructures

Scalable and dynamically balanced shared-everything OLTP with physiological partitioning

Contact Info

Product

Resources

About