• Proceedings of the Korean Information Science Society Conference
• /
• 2000.04a
• /
• pp.92-94
• /
• 2000

자바는 이동성, 호환성, 안정성을 위해서 자바 가상 기계를 이용한 수행 방법을 사용한다. 이는 수행 속도를 저하 시키는 한 요인이다. 그래서, 정적 컴파일 모델과 동적 컴파일 모델과 같은 자바 가상 세계의 성능을 높이기 위한 여러 연구들이 진행되었다. 정적 컴파일 모델은 자바의 특성을 해치며, 동적 컴파일 모델은 초기화 시간이 증가하는 문제점이 있다. 본 논문에서는 디스크 캐쉬 기법을 이용하여 동적 컴파일 모델에서 발생하는 초기화 시간을 줄이는 자바 가상 기계를 구현하였다. SPEC JVM98[1]을 이용하여 Kaffe[2] 와 비교했을 때, 초기화 시간이 2배에서 5배 정도 줄었음을 알 수 있었다.

• Proceedings of the Korean Information Science Society Conference
• /
• 2001.04a
• /
• pp.115-117
• /
• 2001

대부분의 embedded system에서 hard-disk 대용으로 flash memory를 사용하고 있으며, flash device에 압축 커널이미지와 root file system image를 가지고 있다. Booting 고정 중 커널의 압축이 풀리고 메모리에 로드되어 제어를 넘겨받으면 flash memory 상에 존재하는 root file system image를 ramdisk의 image로 로드하여 시스템은 결국 ramdisk에 root file system을 가지게 된다. Ramdisk 상의 프로그램을 실행하기 위해 메모리로 실행파일 이미지를 copy하는 과정을 피하고 ramdisk 상의 이미지를 바로 프로세스의 virtual memory area에 직접 매핑 시켜 주는 XIP(eXection-In-Place)를 구현함으로써 많은 메모리 절감 효과를 얻을 수 있다. 본 연구에서는 ramdisk를 root file system으로 사용하는 embedded system에서의 XIP 구조를 설계하고 구현하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2002.11a
• /
• pp.547-550
• /
• 2002

일반적으로 사용자들은 파일을 효율적으로 관리하기 위해 대용량의 하드 디스크에 다수의 파티션을 생성하여 사용한다. 그러나, 컴퓨터 활용에 익숙치 않은 사용자들의 경우에는 체계적인 파일 관리가 되지 못하여 많은 애로를 겪기도 한다 특히, 하나의 컴퓨터를 여러 사람이 공동으로 사용할 경우에는 체계적인 파일 관리의 부재로 인한 파급 효과가 더욱 클 것이다. 본 논문에서는 기존의 디스크 파티션을 변경하지 않으면서 사용자 입장에서는 개인별로 별도의 디스크 파티션이 부여된 것처럼 느끼도록 하여 파일 관리에 도움을 줄 수 있는 가상 디스크 프로그램을 설계하고 구현하였다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2012.11a
• /
• pp.1599-1602
• /
• 2012

스마트폰이나 태블릿PC와 같은 모바일기기 보급의 확산으로 모바일 클라우드 컴퓨팅이 발전하고 있다. 이와 같은 클라우드 컴퓨팅의 핵심 기술은 가상화 기술이며 서버 가상화가 근간을 이룬다. 가상 서버는 물리 서버의 성능 이상을 추구하고 있으며 디스크 I/O에 따라 성능이 크게 좌우된다. 본 논문에서는 가상 서버 상에서 NAS, Local SAS, PCI-SSD와 같은 다양한 디스크에 대한 I/O 성능을 테스트하였고, 이를 근거로 디스크 I/O 성능에 따른 가상 서버 통합에 대해 고찰하였다.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.20 no.3
• /
• pp.12-26
• /
• 2017

The objective of this study was to develop a runoff analysis system of the Nakdong River watershed using the GRM (Grid-based Rainfall-runoff Model), a physically-based distributed rainfall-runoff model, and to assess the system run time performance according to Microsoft Azure VM (Virtual Machine) settings. Nakdong River watershed was divided into 20 sub-watersheds, and GRM model was constructed for each subwatershed. Runoff analysis of each watershed was calculated in separated CPU process that maintained the upstream and downstream topology. MoLIT (Ministry of Land, Infrastructure and Transport) real-time radar rainfall and dam discharge data were applied to the analysis. Runoff analysis system was run in Azure environment, and simulation results were displayed through web page. Based on this study, the Nakdong River real-time runoff analysis system, which consisted of a real-time data server, calculation node (Azure), and user PC, could be developed. The system performance was more dependent on the CPU than RAM. Disk I/O and calculation bottlenecks could be resolved by distributing disk I/O and calculation processes, respectively, and simulation runtime could thereby be decreased. The study results could be referenced to construct a large watershed runoff analysis system using a distributed model with high resolution spatial and hydrological data.

• The KIPS Transactions:PartD
• /
• v.11D no.1
• /
• pp.11-22
• /
• 2004

As random access memory chip gets cheaper, it becomes affordable to realize main memory-based database systems. Consequently, reducing cache misses emerges as the most important issue in current main memory databases, in which CPU speeds have been increasing at 60% per year, compared to the memory speeds at 10% per you. In this paper, we design and implement a main-memory database system for real-time mobile GIS. Our system is composed of 5 modules： the interface manager provides the interface for PDA users； the memory data manager controls spatial and non-spatial data in main-memory using virtual memory techniques； the query manager processes spatial and non-spatial query : the index manager manages the MR-tree index for spatial data and the T-tree index for non-spatial index : the GIS server interface provides the interface with disk-based GIS. The MR-tree proposed propagates node splits upward only if one of the internal nodes on the insertion path has empty space. Thus, the internal nodes of the MR-tree are almost 100% full. Our experimental study shows that the two-dimensional MR-tree performs search up to 2.4 times faster than the ordinary R-tree. To use virtual memory techniques, the memory data manager uses page tables for spatial data, non- spatial data, T-tree and MR-tree. And, it uses indirect addressing techniques for fast reloading from disk.

• Journal of Korea Multimedia Society
• /
• v.13 no.9
• /
• pp.1299-1313
• /
• 2010

Top-k queries return k objects that users most want in the database. The Partitioned-Layer Index (simply, the PL -index) is a representative method for processing the top-k queries efficiently. The PL-index partitions the database into a number of smaller databases, and then, for each partitioned database, constructs a list of sublayers over the partitioned database. Here, the $i^{th}$ sublayer in the partitioned database has the objects that can be the top-i object in the partitioned one. To retrieve top k results, the PL-index merges the sublayer lists depending on the user's query. The PL-index has the advantage of reading a very small number of objects from the database when processing the queries. However, since many random accesses occur in merging the sublayer lists, query performance of the PL-index is not good in environments like disk-based databases. In this paper, we propose the Abstracted Partitioned-Layer Index (simply, the APL-index) that significantly improves the query performance of the PL-index in disk-based environments by reducing the number of random accesses. First, by abstracting each sublayer of the PL -index into a virtual (point) object, we transform the lists of sublayers into those of virtual objects (ie., the APL-index). Then, we virtually process the given query by using the APL-index and, accordingly, predict sublayers that are to be read when actually processing the query. Next, we read the sublayers predicted from each sublayer list at a time. Accordingly, we reduce the number of random accesses that occur in the PL-index. Experimental results using synthetic and real data sets show that our APL-index proposed can significantly reduce the number of random accesses occurring in the PL-index.

• Journal of KIISE:Computer Systems and Theory
• /
• v.30 no.5_6
• /
• pp.232-238
• /
• 2003

Traditional operating systems use a virtual memory to provide users with a bigger memory than a physical memory. The virtual memory augments the insufficient physical memory by the swap device. Since disks are usually used as the swap device, the cost of a page fault is relatively high compared to the access cost of the physical memory. Recently, numerous papers have investigated the Network RAM in order to exploit the idle memory in the network instead of disks. Since today's distributed systems are interconnected with high-performance networks, the network latency is far smaller than the disk access latency In this paper we design and implement the Network RAM using block device driver on Linux. This is the first implementation of the Network RAM on Linux. We propose the new reliability method to recover the page when the other workstation's memory is damaged. The system using the Network RAM as the swap device reduces the execution time by 40.3% than the system using the disk as the swap device. The performance results suggest that the new reliability method that use the processor more efficiently has the similar execution time with others, but uses smaller server memory and generates less message traffic than others.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.28 no.2
• /
• pp.429-436
• /
• 2018

Electronic documents are efficient to be created and managed, but they are liable to lose their originality because copies are created during distribution and delivery. For this reason, various security technologies for electronic documents have been applied. However, most security technologies currently used are for document management such as file access privilege control, file version and history management, and therefore can not be used in environments where authenticity is absolutely required, such as confidential documents. In this paper, we propose a method to detect document forgery and tampering through analysis of file system without installing an agent inside the instance operating system in cloud computing environment. BubbleDoc monitors the minimum amount of virtual volume storage in an instance, so it can efficiently detect forgery and tampering of documents. Experimental results show that the proposed technique has 0.16% disk read operation overhead when it is set to 1,000ms cycle for monitoring for document falsification and modulation detection.

• Journal of KIISE:Computer Systems and Theory
• /
• v.36 no.2
• /
• pp.102-110
• /
• 2009

The memory wall is the growing disparity of speed between CPU and memory outside the CPU chip. An economical solution is a memory hierarchy organized into several levels, such as processor registers, cache, main memory, disk storage. We introduce a novel memory hierarchy optimization technique in Linux based embedded systems using on-chip SRAM for the first time. The optimization technique allocates On-Chip SRAM to the code/data that selected by programmers by using virtual memory systems. Experiments performed with nine applications indicate that the runtime improvements can be achieved by up to 35%, with an average of 14%, and the energy consumption can be reduced by up to 40%, with an average of 15%.