Computer Science > Distributed, Parallel, and Cluster Computing
[Submitted on 28 Aug 2022]
Title:Reshape: Adaptive Result-aware Skew Handling for Exploratory Analysis on Big Data
View PDFAbstract:The process of data analysis, especially in GUI-based analytics systems, is highly exploratory. The user iteratively refines a workflow multiple times before arriving at the final workflow. In such an exploratory setting, it is valuable to the user if the initial results of the workflow are representative of the final answers so that the user can refine the workflow without waiting for the completion of its execution. Partitioning skew may lead to the production of misleading initial results during the execution. In this paper, we explore skew and its mitigation strategies from the perspective of the results shown to the user. We present a novel framework called Reshape that can adaptively handle partitioning skew in pipelined execution. Reshape employs a two-phase approach that transfers load in a fine-tuned manner to mitigate skew iteratively during execution, thus enabling it to handle changes in input-data distribution. Reshape has the ability to adaptively adjust skew-handling parameters, which reduces the technical burden on the users. Reshape supports a variety of operators such as HashJoin, Group-by, and Sort. We implemented Reshape on top of two big data engines, namely Amber and Flink, to demonstrate its generality and efficiency, and report an experimental evaluation using real and synthetic datasets.
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.