• Proceedings of the Korean Information Science Society Conference
• /
• 2012.06a
• /
• pp.218-220
• /
• 2012

인터넷 활용과 웹 어플리케이션의 개발이 증가함에 따라 처리해야하는 데이터의 양도 또한 증가하고 있다. 대량의 데이터를 효과적으로 처리하기 위한 방법 중 하나로 병렬처리 프로그래밍 모델인 맵리듀스가 있다. 하둡은 맵리듀스의 오픈소스 구현으로 대량의 데이터를 병렬로 처리하는 무료 자바 소프트웨어 프레임워크이다. 분산 파일 시스템을 사용하는 하둡에서는 처리하는 데이터가 다른 노드에 위치하는 데이터 로컬리티 문제가 전체 작업 수행시간의 증가를 야기하는 문제가 있다. 본 논문에서는 하둡에서의 데이터 로컬리티 문제를 해결하기 위한 데이터 복제기법을 제안한다. 제안하는 데이터 복제기법에서는 1) 라그랑지 보간법을 사용하여 과거 접근수를 이용한 미래 접근수를 예측하고, 2) 예측된 값을 Threshold값으로 설정하고, 3) 데이터 로컬리티 문제가 발생하였을 때, 복제사본을 생성할 것인지 캐시를 생성할 것인지를 결정하여 복제 사본의 수를 최적화 한다. 실험을 통해 단순히 복제사본 수를 증가시킴으로써 데이터 로컬리티를 향상을 이루어도 작업 완료시간이 감소하는 것이 아니라는 결과를 볼 수 있었고, 오버 런치로 인한 작업 완료시간 증가를 줄이기 위해 데이터 복제사본 수 최적화의 필요성을 확인할 수 있었다.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.15 no.1
• /
• pp.107-114
• /
• 2020

This paper studies a method for estimating the scale of a Hadoop cluster to process big data as a cost-effective manner. In the case of medical institutions, demands for cloud-based big data analysis are increasing as medical records can be stored outside the hospital. This paper first analyze the Amazon EMR framework, which is one of the popular cloud-based big data framework. Then, this paper presents a efficiency model for scaling the Hadoop cluster to execute a Mapreduce application more cost-effectively. This paper also analyzes the factors that influence the execution of the Mapreduce application by performing several experiments under various conditions. The cost efficiency of the analysis of the big data can be increased by setting the scale of cluster with the most efficient processing time compared to the operational cost.

• Journal of Digital Contents Society
• /
• v.15 no.3
• /
• pp.365-371
• /
• 2014

In the era of big data, algorithms for the existing IT environment cannot accept on a distributed architecture such as hadoop. Thus, new distributed algorithms which apply a distributed framework such as MapReduce are needed. Lloyd's algorithm commonly used for vector quantization is developed using MapReduce recently. In this paper, we proposed a decombined distributed VQ codebook generation algorithm based on a distributed VQ codebook generation algorithm using MapReduce to get a result more fast. The result of applying the proposed algorithm to big data showed higher performance than the conventional method.

• Journal of Digital Contents Society
• /
• v.16 no.4
• /
• pp.545-553
• /
• 2015

While all of the data has a value in itself, most of the data that is collected in the real world is a random and unstructured. In order to extract useful information from the data, it is need to use the data transform and analysis algorithms. Data mining is used for this purpose. Today, there is not only need for a variety of data mining techniques to analyze the data but also need for a computational requirements and rapid analysis time for huge volume of data. The method commonly used to store huge volume of data is to use the hadoop. A method for analyzing data in hadoop is to use the MapReduce framework. In this paper, we developed a tournament-based MapReduce method for high efficiency in developing an algorithm on a single machine to the MapReduce framework. This proposed method can apply many analysis algorithms and we showed the usefulness of proposed tournament based method to apply frequently used data mining algorithms k-means and k-nearest neighbor classification.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2016.07a
• /
• pp.191-193
• /
• 2016

본 논문에서는 SSD 저장장치를 포함하는 하둡의 Iterative Processing에 대한 성능 분석 결과를 소개한다. 하둡은 맵 리듀스 병렬 프로그래밍 모델을 통해 Batch Processing에 특화된 구조를 가지고 있는 프레임 워크이다. 이는 병렬/분산 환경에서 큰 성능향상을 보장하지만, 반복 작업을 수행하는 Iterative Processing에 대하여는 성능이 낮아지는 문제가 존재하고 있다. 이에 본 논문에서는 점차 낮아지는 가격으로 인해 하둡시스템에 적용 가능성이 타진되는 SSD를 통해 반복 작업의 성능이슈를 해결할 수 있는지 확인하고, SSD를 통한 성능향상의 요소가 존재하는지 알아보고자 실험을 진행하였다. 실험에서는 Batch Processing인 word count와 Iterative Processing인 Page Rank 알고리즘을 MapReduce로 구현하고 데이터 크기에 따른 성능 향상도를 측정하였고, SSD 추가와 같은 하드웨어적인 성능을 통한 하둡의 반복 작업은 큰 효율을 기대하기가 어렵다는 결론을 보였다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2016.05a
• /
• pp.376-377
• /
• 2016

Performance of Hadoop, which is a distributed data processing framework for big data analysis, is affected by several characteristics of each node in distributed cluster such as processing power and network bandwidth. This paper analyzes previous approaches for heterogeneous hadoop clusters, and presents several requirements for distributed node clustering in small-and-medium sized hospitals by considering computing environments of the hospitals.

• KIPS Transactions on Software and Data Engineering
• /
• v.3 no.8
• /
• pp.293-298
• /
• 2014

Detecting and understanding the changes between RDF data is crucial in the evolutionary process, synchronization system, and versioning system on the web of data. However, current researches on detecting changes still remain unsatisfactory in that they did neither consider the large scale of RDF data nor accurately produce the RDF deltas. In this paper, we propose a scalable and effective change detection using a MapReduce framework which has been used in many fields to process and analyze large volumes of data. In particular, we focus on the structure-based change detection that adopts a strategy for the comparison of blank nodes in RDF data. To achieve this, we employ a method which is composed of two MapReduce jobs. First job partitions the triples with blank nodes by grouping each triple with the same blank node ID and then computes the incoming path to the blank node. Second job partitions the triples with the same path and matchs blank nodes with the Hungarian method. In experiments, we show that our approach is more accurate and effective than the previous approach.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2014.05a
• /
• pp.479-482
• /
• 2014

Nowadays open-source hadoop systems have been using widely to efficiently manage a fast-growing big data. Hadoop systems consist of distributed file processing system called HDFS (Hadoop Distributed File System) and distributed parallel processing system called MapReduce. The MapReduce reads and processes big data from HDFS and then processed results are written in HDFS again by the MapReduce. Such a processing method has different system structure respectively according to hadoop version. Therefore, this paper shows analysis results for performance of hadoop systems. For this, we devise a way which monitors hadoop systems and measure occurrence frequency of processes, threads, and variables generated in hadoop system itself using the devised way. So, by using the measured results as analysis indicator, we help the indicator predict inner performance of hadoop systems.

• Journal of Internet Computing and Services
• /
• v.20 no.1
• /
• pp.87-96
• /
• 2019

In this paper, we propose a MapReduce-supported clustering technique for collecting and classifying distributed workflow enactment event logs as a preprocessing tool. Especially, we would call the distributed workflow enactment event logs as Workflow BIG-Logs, because they are satisfied with as well as well-fitted to the 5V properties of BIG-Data like Volume, Velocity, Variety, Veracity and Value. The clustering technique we develop in this paper is intentionally devised for the preprocessing phase of a specific workflow process mining and analysis algorithm based upon the workflow BIG-Logs. In other words, It uses the Map-Reduce framework as a Workflow BIG-Logs processing platform, it supports the IEEE XES standard data format, and it is eventually dedicated for the preprocessing phase of the ${\rho}$-Algorithm that is a typical workflow process mining algorithm based on the structured information control nets. More precisely, The Workflow BIG-Logs can be classified into two types: of activity-based clustering patterns and performer-based clustering patterns, and we try to implement an activity-based clustering pattern algorithm based upon the Map-Reduce framework. Finally, we try to verify the proposed clustering technique by carrying out an experimental study on the workflow enactment event log dataset released by the BPI Challenges.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.9
• /
• pp.1731-1737
• /
• 2017

K Nearest Neighbor Join (KNN Join) is a simple yet effective method in machine learning. It is widely used in small dataset of the past time. As the number of data increases, it is infeasible to run this model on an actual application by a single machine due to memory and time restrictions. Nowadays a popular batch process model called MapReduce which can run on a cluster with a large number of computers is widely used for large-scale data processing. Hadoop is a framework to implement MapReduce, but its performance can be further improved by a new framework named Spark. In the present study, we will provide a KNN Join implement based on Spark. With the advantage of its in-memory calculation capability, it will be faster and more effective than Hadoop. In our experiments, we study the influence of different factors on running time and demonstrate robustness and efficiency of our approach.