Dynamic load balancing in very large shared-nothing hypercube database computers

Hua, Kien A.; Su, Jonathan

doi:10.1109/12.260633

Cited by 7 publications

(3 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Their works researched four types of partition skew and modeled corresponding data skew handling strategies. Hua and Lee [6] handled data skew in multiprocessor database systems using partition tuning, and Hua and Su [7] considered the causes and characteristics of data skew and described a dynamic data balancing method in very large shared nothing hypercube database systems. Lu and Yu [8] researched a full dynamic partitioning approach faced to data skew issues that can effectively distribute load among processing nodes without priori knowledge of data distribution.…”

Section: Related Workmentioning

confidence: 99%

A data skew handling method based on the minimum spatial proximity for parallel spatial database

2009

View full text Add to dashboard Cite

Data skew is one of most important reasons to deteriorate the performance of parallel spatial database. This paper studies the issues of handling data skew in shared nothing parallel spatial database system architecture. A novel data skew handling method is proposed, which fulfill spatial data distribution balancing based on the spatial proximity of data fragments. The minimum spatial proximity is used to be the principle of moving data fragments among different network parallel nodes. Our experimental results show that the proposed data skew handling method can achieve dynamic data load balancing and offer significant improvement for reducing response time of parallel spatial queries.

show abstract

Section: Related Workmentioning

confidence: 99%

A data skew handling method based on the minimum spatial proximity for parallel spatial database

2009

View full text Add to dashboard Cite

show abstract

“…Reference [7] dealt with data unbalancing issue using partition tuning ways. Reference [8] considered the causes and characteristics of data skew and described a dynamic data balancing method in very large shared nothing hypercube database systems. Reference [9] researched a full dynamic partitioning approach faced to data skew issues that can effectively distribute load among processing nodes without priori knowledge of data distribution, and [10] provided NBRADJUST and REORDER data balancing methods.…”

Section: Introductionmentioning

confidence: 99%

Spatial Data Dynamic Balancing Distribution Method Based on the Minimum Spatial Proximity for Parallel Spatial Database

Zhou¹,

Zhu²,

Zhang³

2011

JSW

View full text Add to dashboard Cite

Spatial data balancing distribution can evidently improve the performance of parallel spatial database in shared nothing parallel architecture. Considering spatial locality and unstructured variable length characteristics of spatial data, this paper proposes a dynamic spatial data balancing distribution method under shared nothing parallel database environment. By using Hilbert ordering code to keep spatial locality relationship between spatial objects, the presented method can fulfill spatial data static balancing distribution status, which depends on hierarchically decomposing Hilbert space-filling curve code to allocate approximately even spatial data volume to parallel nodes in distributed network. Then spatial proximity index is introduced to resolve spatial data unbalancing problem caused by spatial database dynamic updating. Through moving spatial data fragments to goal node which is the minimum spatial proximity with data moving out node, the spatial database redistributing strategy can attain dynamic data balance among all network parallel nodes. Our experimental results show that the proposed method can effectively improve parallel performance deterioration resulted from spatial data unbalancing and achieve spatial data dynamic balancing distribution in parallel spatial database

show abstract

“…Skewed load distribution extends execution time to the finish time of the most heavily loaded PE. Thus, there has been much research on balancing loads among PEs for parallel database systems [e.g., 2,6,7,10,12].…”

Section: Introductionmentioning

confidence: 99%

Directory‐traverse‐cost‐based skew handling for parallel data access

Watanabe

Yokota

2003

Systems & Computers in Japan

View full text Add to dashboard Cite

SUMMARYTechniques for load balancing by adjusting location of data are widely studied, because they are quite influential in improving the data-access performance and the scalability of a parallel system. To make load balancing effective, distributed directory structures, methods for evaluating loads, and distributed control mechanism for handling skews are important. In this paper, we propose a distributed algorithm to evaluate loads precisely. It counts the loads of intermediate index nodes of a distributed directory, which were ignored in previous publications. We also propose a linear-order distributed skew-handling algorithm adopting the proposed load evaluation algorithm. We validate the effect of our load evaluation algorithm through simulations.

show abstract

Dynamic load balancing in very large shared-nothing hypercube database computers

Cited by 7 publications

References 23 publications

A data skew handling method based on the minimum spatial proximity for parallel spatial database

A data skew handling method based on the minimum spatial proximity for parallel spatial database

Spatial Data Dynamic Balancing Distribution Method Based on the Minimum Spatial Proximity for Parallel Spatial Database

Directory‐traverse‐cost‐based skew handling for parallel data access

Contact Info

Product

Resources

About