Adaptive and big data scale parallel execution in oracle

Bellamkonda, Srikanth; Li, Huagang; Jagtap, Unmesh; Zhu, Yan; Liang, Vince; Cruanes, Thierry

doi:10.14778/2536222.2536235

Cited by 28 publications

(17 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Kim [10] and Morzy [11] described the transformation of logical sets (magic set), which leads to new insights into the costly units' request, thus allowing the volume of data being processed to be reduced.…”

Section: Related Workmentioning

confidence: 99%

Development of Query Transformation Method by Cost Optimization

Altayeva

Yong-Jin

Cho

2016

IJFIS

View full text Add to dashboard Cite

The transformation time among queries in the database management system (DBMS) is responsible for the execution time of users' queries, because a conventional DBMS does not consider the transformation cost when queries are transformed for execution. To reduce the transformation time (cost reduction) during execution, we propose an optimal query transformation method by exploring queries from a cost-based point of view. This cost-based point of view means considering the cost whenever queries are transformed for execution. Toward that end, we explore and compare set off heuristic, linear, and exhaustive costbased transformations. Further, we describe practical methods of cost-based transformation integration and some query transformation problems. Our results show that, some cost-based transformations significantly improve query execution time. For instance, linear and heuristic transformed queries work 43% and 74% better than exhaustive queries.

show abstract

Section: Related Workmentioning

confidence: 99%

Development of Query Transformation Method by Cost Optimization

Altayeva

Yong-Jin

Cho

2016

IJFIS

View full text Add to dashboard Cite

show abstract

“…The 64 available hardware threads are distributed uniformly over the streams, and each stream executes random permutations of the TPC-H queries. Figure 12 shows that the throughput stays high 6 In practice, the database itself is located on a single NUMA node, because the data is read from disk by a single thread. Other allocations are local to the thread that first wrote to that memory.…”

Section: Elasticitymentioning

confidence: 99%

“…Very similar to Volcano-style parallelization, in Oracle the individual operators are largely unaware of parallelism. [6] addresses some problems of this model, in particular reliance on query optimizer estimates, by adaptively changing data distribution decisions during query execution. In an experimental study Kiefer et al [17] showed that NUMA-awareness can improve database performance considerably.…”

Section: Related Workmentioning

confidence: 99%

Morsel-driven parallelism

Leis

Boncz

Kemper

et al. 2014

Proceedings of the 2014 ACM SIGMOD International Conference on Management of Data

187

View full text Add to dashboard Cite

With modern computer architecture evolving, two problems conspire against the state-of-the-art approaches in parallel query execution: (i) to take advantage of many-cores, all query work must be distributed evenly among (soon) hundreds of threads in order to achieve good speedup, yet (ii) dividing the work evenly is difficult even with accurate data statistics due to the complexity of modern out-of-order cores. As a result, the existing approaches for "plandriven" parallelism run into load balancing and context-switching bottlenecks, and therefore no longer scale. A third problem faced by many-core architectures is the decentralization of memory controllers, which leads to Non-Uniform Memory Access (NUMA).In response, we present the "morsel-driven" query execution framework, where scheduling becomes a fine-grained run-time task that is NUMA-aware. Morsel-driven query processing takes small fragments of input data ("morsels") and schedules these to worker threads that run entire operator pipelines until the next pipeline breaker. The degree of parallelism is not baked into the plan but can elastically change during query execution, so the dispatcher can react to execution speed of different morsels but also adjust resources dynamically in response to newly arriving queries in the workload. Further, the dispatcher is aware of data locality of the NUMA-local morsels and operator state, such that the great majority of executions takes place on NUMA-local memory. Our evaluation on the TPC-H and SSB benchmarks shows extremely high absolute performance and an average speedup of over 30 with 32 cores.

show abstract

“…Traditionally, in relational DBMS, window functions have been commonly used for data analytics [7,1]. Instead of performing analysis (e.g.…”

Section: Introductionmentioning

confidence: 99%

Towards Window Analytics over Large-scale Graphs

Fan

Tan

2015

Proceedings of the 2015 ACM SIGMOD on PhD Symposium

View full text Add to dashboard Cite

In relational DBMS, window functions have been widely used to facilitate data analytics. Surprisingly, while similar concepts have been employed for graph analytics, there has been no explicit notions of graph window analytic functions. In this paper, we formally introduce window queries for graph analytics. In such queries, for each vertex, the analysis is performed on a window of vertices defined based on the graph structure. In particular, we identify three instantiations, namely the unified window, the k-hop window and the topological window. We focus on processing the latter two window queries and develop two novel indices, Dense Block index (DBIndex) and Inheritance index (I-Index), to facilitate efficient processing of these two types of windows respectively.

show abstract

Adaptive and big data scale parallel execution in oracle

Cited by 28 publications

References 13 publications

Development of Query Transformation Method by Cost Optimization

Development of Query Transformation Method by Cost Optimization

Morsel-driven parallelism

Towards Window Analytics over Large-scale Graphs

Contact Info

Product

Resources

About