• 한국콘텐츠학회논문지
• /
• 제14권5호
• /
• pp.8-17
• /
• 2014

슈퍼컴퓨팅 자원들은 주로 MPI와 같은 메시지 교환 인터페이스에 기반한 통신 집적도가 높은 고성능 컴퓨팅(HPC: High Performance Computing) 응용 분야를 지원하는데 활용되어 왔다. 반면에, 대규모 계산처리 컴퓨팅(HTC: High Throughput Computing) 방식의 패러다임은 주로 계산 집적도가 높고(상대적으로 적은 I/O 연산), 독립적인(작업들 간의 통신이 적음) 많은 수의 작업을 처리하는 것을 요구하고 있다. 국내에서도 고에너지 물리, 신약개발, 핵물리와 같은 연구 분야를 중심으로 대규모 컴퓨팅 자원을 요구하는 계산처리에 대한 수요가 증가하고 있다. 본 논문에서는 이러한 HTC 과학 응용들에 대한 효율적인 지원을 국가차원의 슈퍼컴퓨팅 분산 환경에서 제공하기 위해 연구/개발되어진 대규모 계산처리 서비스(HTCaaS: High Throughput Computing as a Service)의 전체 구조 및 구성 요소, 실행 시나리오 및 실제 응용 적용 사례 등에 대해 서술한다.

• 한국전산구조공학회:학술대회논문집
• /
• 한국전산구조공학회 2007년도 정기 학술대회 논문집
• /
• pp.719-724
• /
• 2007

Many engineering problems often require the large amount of computing resources for iterative simulations of problems treating many parameters and input files. In order to overcome the situation, this paper proposes an e-Science based computational system. The system exploits the Grid computing technology to establish an integrated web service environment which supports distributed high throughput computational simulations and remote executions. The proposed system provides an easy-to-use parametric study service where a computational service includes real time monitoring. To verify usability of the proposed system, two kinds of applications were introduced. The first application is an Aerospace Integrated Research System (e-AIRS). The e-AIRS adapts the proposed computational system to solve CFD problems. The second one is design and optimization of protein 3-dimensional structures.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권7호
• /
• pp.2977-2997
• /
• 2018

In big data era, fresh data grows rapidly every day. More than 30,000 gigabytes of data are created every second and the rate is accelerating. Many organizations rely heavily on real time streaming, while big data stream computing helps them spot opportunities and risks from real time big data. Storm, one of the most common online stream computing platforms, has been used for big data stream computing, with response time ranging from milliseconds to sub-seconds. The performance of Storm plays a crucial role in different application scenarios, however, few studies were conducted to evaluate the performance of Storm. In this paper, we investigate the performance of Storm under different application scenarios. Our experimental results show that throughput and latency of Storm are greatly affected by the number of instances of each vertex in task topology, and the number of available resources in data center. The fault-tolerant mechanism of Storm works well in most big data stream computing environments. As a result, it is suggested that a dynamic topology, an elastic scheduling framework, and a memory based fault-tolerant mechanism are necessary for providing high throughput and low latency services on Storm platform.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2017년도 춘계학술발표대회
• /
• pp.171-173
• /
• 2017

최근 모바일 컴퓨팅은 어플리케이션 실행, 이미지 처리, 동영상 인코딩, 게임 등의 모바일 컴퓨팅 작업 처리를 위한 높은 성능의 컴퓨팅 능력을 요구한다. 모바일 컴퓨팅의 성능을 향상하기 위해 모바일 클라우드 컴퓨팅(Mobile Cloud Computing)을 도입하였다. 기존 모바일 클라우드 컴퓨팅에서는 모바일 컴퓨팅 작업의 처리를 위해 고-성능 컴퓨팅(High-Performance Computing)방법을 적용한 오프로드가 연구되고 있다. 고-성능 컴퓨팅의 목적은 단일작업의 처리속도 향상이므로 다중 작업처리를 위한 모바일 클라우드 컴퓨팅에는 적합하지 않다. 또한 고-성능 컴퓨팅은 모바일 클라우드 사용자에게 동등한 컴퓨팅 성능을 제공하지 못하는 문제점을 내재한다. 본 논문에서는 모바일 클라우드 컴퓨팅 기반 다중 작업 처리를 위한 Multitasking Mechanism for High-Throughput Computing(M2-HTC)을 제안한다. M2-HTC는 모바일 클라우드에서 처리중인 작업과 사용 가능한 컴퓨팅 리소스를 활용하여 다중 모바일 컴퓨팅 작업의 처리시간을 최소화하는 기법이다.

• 한국전산구조공학회논문집
• /
• 제20권3호
• /
• pp.365-370
• /
• 2007

공학 및 과학문제 해석을 위해 적용되는 전산 시뮬레이션은 다양한 변수 혹은 데이터의 변화를 통해 다수의 작업을 생성하고 계산함과 동시에, 생성된 결과를 비교 분석하기 위한 필수적인 기법이다. 본 연구에서는 그리드 컴퓨팅을 활용하여 웹상에서 대용량의 전산 시뮬레이션이 가능한 시스템을 개발하고, 이를 이용한 2가지 실제 응용사례를 제시한다. 첫 번째 응용사례는 e-AIRS(Aerospace Integrated Research Environment)라 명명된 연구포탈이다 e-AIRS는 수치해석 연구자가 대규모의 전산 해석을 실시하고, 실험 연구자가 원격지에서 실험을 요청하고 그 결과를 모니터링 할 수 있는 e-Science 연구환경을 제공한다. 두 번째 응용사례는 대규모 계산환경을 이용한 단백질 구조설계를 제시한다. 제안된 계산환경을 이용하여 생성된 단백질 전산 예측구조와 자연상태 구조를 비교하고, 제안된 계산환경의 유용성을 검토한다.

• 정보과학회 논문지
• /
• 제42권5호
• /
• pp.573-579
• /
• 2015

국가 슈퍼컴퓨팅 인프라는 국내 여러 지역에 분산된 슈퍼컴퓨팅 클러스터들로 이루어져 있으며, 본 연구팀에서는 이러한 이기종의 지리적으로 분산된 클러스터들을 대규모 과학 응용 연구자들에게 효율적으로 제공하기 위해 대규모 계산처리 시스템인 HTCaaS(High-Throughput Computing as a Service)를 자체 개발하였다. 본 논문에서는 이러한 대규모 계산처리 시스템(HTCaaS)을 활용하여 각 계산 자원을 동적으로 관리하는 방법에 대해서 논의하고, 신약재창출이라는 실제 과학 응용을 통해 그 효율성을 검증한다. 특히 유효 자원 식별을 위한 대기시간 및 성공률 개념을 이용한 동적 계산 자원 관리 기술을 적용함으로써 자원 활용률과 정확성, 신뢰성, 편의성이 향상될 수 있으며, 그 결과 전체적인 작업 시간의 단축과 작업 처리량도 향상될 수 있음을 확인할 수 있었다.

• IEIE Transactions on Smart Processing and Computing
• /
• 제3권5호
• /
• pp.275-284
• /
• 2014

This paper considers two-tier networks consisting of macrocells and femtocells operating in the same spectrum. This paper proposes a femtocell spectrum reuse scheme that determines the shared spectrum and transmit power for the femtocells to mitigate the effects of cross-tier interference between the macrocells and femtocells. The proposed scheme provides macrocell throughput that is unaffected by the increasing number of femtocells per cell site and improves the femtocell signal quality at the same time by limiting the cross-tier interference. This study analyzed the per-tier signal-to-interference ratio (SIR) and outage probability of the proposed scheme to investigate the macrocell and femtocell performance. The total throughput of the proposed scheme was analyzed based on the outage probabilities. The analysis and numerical results proved that high femtocell throughput can be achieved using only a small fraction of the spectrum while protecting the macrocell throughput. As a result, an improved total throughput was achieved enforcing higher spatial reuse.

• International Journal of Contents
• /
• 제12권2호
• /
• pp.6-11
• /
• 2016

We developed a pilot-job based large-scale distributed computing system to support HTC and MTC, called HTCaaS (High-Throughput Computing as a Service), which helps scientists solve large-scale scientific problems in areas such as pharmaceutical domains, high-energy physics, nuclear physics and bio science. Since most of these problems involve critical data that affect the national economy and activate basic industries, data privacy is a very important issue. In this paper, we implement a privacy enhanced data security mechanism to support HTC and MTC in a large-scale distributed computing system and show how this technique affects performance in our system. With this mechanism, users can securely store data in our system.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제14권12호
• /
• pp.4648-4663
• /
• 2020

This paper proposes the Lempel-Ziv 4(LZ4) compression accelerator optimized for scale-out servers in data centers. In order to reduce CPU loads caused by compression, we propose an accelerator solution and implement the accelerator on an Field Programmable Gate Array(FPGA) as heterogeneous computing. The LZ4 compression hardware accelerator is a fully pipelined architecture and applies 16 dictionaries to enhance the parallelism for high throughput compressor. Our hardware accelerator is based on the 20-stage pipeline and dictionary architecture, highly customized to LZ4 compression algorithm and parallel hardware implementation. Proposing dictionary architecture allows achieving high throughput by comparing input sequences in multiple dictionaries simultaneously compared to a single dictionary. The experimental results provide the high throughput with intensively optimized in the FPGA. Additionally, we compare our implementation to CPU implementation results of LZ4 to provide insights on FPGA-based data centers. The proposed accelerator achieves the compression throughput of 639MB/s with fine parallelism to be deployed into scale-out servers. This approach enables the low power Intel Atom processor to realize the Hadoop storage along with the compression accelerator.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권2호
• /
• pp.974-987
• /
• 2018

This paper presents a novel and efficient data transmission scheme based on mobile edge computing for the massive IoT environments which should support various type of services and devices. Based on an accurate and precise synchronization process, it maximizes data transmission throughput, and consistently maintains a flow's latency. To this end, the proposed efficient software defined data transmission scheme (ESD-DTS) configures and utilizes synchronization zones in accordance with the 4 usage cases, which are end node-to-end node (EN-EN), end node-to-cloud network (EN-CN), end node-to-Internet node (EN-IN), and edge node-to-core node (EdN-CN); and it transmit the data by the required service attributes, which are divided into 3 groups (low-end group, medium-end group, and high-end group). In addition, the ESD-DTS provides a specific data transmission method, which is operated by a buffer threshold value, for the low-end group, and it effectively accommodates massive IT devices. By doing this, the proposed scheme not only supports a high, medium, and low quality of service, but also is complied with various 5G usage scenarios. The essential difference between the previous and the proposed scheme is that the existing schemes are used to handle each packet only to provide high quality and bandwidth, whereas the proposed scheme introduces synchronization zones for various type of services to manage the efficiency of each service flow. Performance evaluations show that the proposed scheme outperforms the previous schemes in terms of throughput, control message overhead, and latency. Therefore, the proposed ESD-DTS is very suitable for upcoming 5G networks in a variety of massive IoT environments with supporting mobile edge computing (MEC).