• Journal of KIISE
• /
• v.43 no.1
• /
• pp.27-33
• /
• 2016

Virtualization has become one of the most helpful techniques in computing systems, and today it is prevalent in several computing environments including desktops, data-centers, and enterprises. However, since I/O layers are implemented to be oblivious to the I/O behaviors on virtual machines (VM), there still exists an I/O scalability issue in virtualized systems. In particular, when a multi-queue solid state drive (SSD) is used as a secondary storage, each system reveals a semantic gap that degrades the overall performance of the VM. This is due to two key problems, accelerated lock contentions and the I/O parallelism issue. In this paper, we propose a novel approach, including the design of virtual CPU (vCPU)-dedicated queues and I/O threads, which efficiently distributes the lock contentions and addresses the parallelism issue of Virtio-blk-data-plane in virtualized environments. Our approach is based on the above principle, which allocates a dedicated queue and an I/O thread for each vCPU to reduce the semantic gap. Our experimental results with various I/O traces clearly show that our design improves the I/O operations per second (IOPS) in virtualized environments by up to 155% over existing QEMU-based systems.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.16 no.1
• /
• pp.109-116
• /
• 2021

Since RDF (Resource Description Framework) triples are modeled as graph, we cannot directly adopt existing solutions in relational databases and XML technology. In order to store, index, and query Linked Data more efficiently, we propose a convergence indexing method combined R*-tree and K-dimensional trees. This method uses a hybrid storage system based on HDD (Hard Disk Drive) and SSD (Solid State Drive) devices, and a separated filter and refinement index structure to filter unnecessary data and further refine the immediate result. We perform performance comparisons based on three standard join retrieval algorithms. The experimental results demonstrate that our method has achieved remarkable performance compared to other existing methods such as Quad and Darq.

• Journal of KIISE:Databases
• /
• v.35 no.5
• /
• pp.400-410
• /
• 2008

Flash memory has been widely used in mobile devices, such as mobile phone and digital camera. Recently flash SSD(Solid State Disk), having same interface of the disk drive, is replacing the hard disk of some laptop computers. However, flash memory still cannot be considered as the storage of database systems. The FTL(Flash Translation Layer) of commercial flash SSD, making flash memory operate exactly same as a hard disk, shows poor performance on the workload of databases with many random overwrites. Recently In-Page Logging(IPL) approach was proposed to solve this problem. In this paper, we implement IPL approach on SQLite, a popular open source embedded DBMS, and evaluate its performance. It improves the performance by up to 30 factors for update queries.

• Annual Conference of KIPS
• /
• 2012.04a
• /
• pp.1121-1124
• /
• 2012

열-지향 데이터베이스 시스템인 C-Store는 많은 상용 데이터베이스 시스템과는 달리 데이터를 행(row) 위주가 아닌 열(column) 위주로 저장을 하여, 데이터 웨어하우스와 같이 주로 읽기 IO를 유발하는 환경에서 데이터의 전송량을 줄임으로써, 높은 성능을 보였다. 본 논문에서는 대표적인 열 지향 저장 DBMS인 C-Store와 행 위주의 저장구조를 사용하는 기존 DBMS와의 차이점을 알아보고, C-Store의 저장장치로 하드디스크와 차세대 저장장치로 주목받고 있는 플래시 SSD(Solid State Disk)를 사용하였을 때, 발생할 수 있는 장단점에 대해 분석하였다.

• Annual Conference of KIPS
• /
• 2010.11a
• /
• pp.1777-1779
• /
• 2010

최근 저전력과 새로운 응용의 대용량 데이터 처리 요구에 따라 저장 장치로 하드디스크 대신 빠른 입출력 성능을 가진 SSD 를 저장장치로 활용하는 사례가 증가하고 있으나 빈번한 임의 쓰기를 발생하는 소규모 특정 데이터를 하드디스크나 SSD 에 저장하는 경우 발생하는 성능 및 안정성 저하문제는 아직 완전히 해결하지 못하고 있다. 본 논문에서는 빈번한 임의 쓰기를 요구하는 인덱스를 바이트 접근이 가능한 NVRAM 에 저장하는 시스템 구조를 제안한다. 제안하는 시스템 구조는 NVRAM 의 바이트 단위 빠른 읽기/쓰기와 인덱스 페이지 내 실제 데이터 변경 크기가 블록 크기보다 훨씬 작다는 특성을 고려하여 인덱스 페이지 크기를 최적화하고 메인 메모리에서 인덱스 페이지 버퍼를 따로 할당하여 관리함으로써 시스템 성능 향상을 기대할 수 있음을 보인다.

• JSTS:Journal of Semiconductor Technology and Science
• /
• v.15 no.5
• /
• pp.570-576
• /
• 2015

Existing flash storage devices such as universal flash storage and solid state disk support command queuing to improve storage I/O bandwidth. Command queuing allows multiple read/write requests to be pending in a device queue. Because multi-channel and multi-way architecture of flash storage devices can handle multiple requests simultaneously, command queuing is an indispensable technique for utilizing parallel architecture. However, command queuing can be harmful to the latency of fsync system call, which is critical to application responsiveness. We propose a dynamic queue depth adaptation technique, which reduces the queue depth if user application is expected to send fsync calls. Experiments show that the proposed technique reduces the fsync latency by 79% on average compared to the original scheme.

• Journal of the Korea Convergence Society
• /
• v.4 no.1
• /
• pp.21-26
• /
• 2013

Due to a lack of research regarding disruptive technologies in domestic research, the purpose of this study is to aid in the understanding of disruptive technologies through empirical analysis of cases selected in the computer data storage industry. Analysis results have shown that SSDs, which threaten the existence of HDDs, adhere to the conditions of being a disruptive technology as first presented by Christensen(1992). SSDs are not only technologically superior to HDDs but can be mass produced due to its applicability in a vast array of product categories made possible by their miniaturization, weight reduction, and safety. This diversity of applicable fields makes it possible for mass production leading to further decrease in the unit price ultimately continuing the diffusion of this technology. By presenting empirical cases to aid in the understanding of disruptive technology, it is determined that the findings of this study contribute greatly to both academia and the business world.

• Journal of KIISE
• /
• v.42 no.2
• /
• pp.169-176
• /
• 2015

The atomic write technique is a good solution to solve the problem of the double write buffer. The atomic write technique needs modified I/O subsystems (i.e., file system and I/O schedulers) and a special SSD that guarantees the atomicity of the write request. In this paper, we propose the writing unit aligned block allocation technique (for EXT4 file system) and the merge prevention of requests technique for the CFQ scheduler. We also propose an atomic write-supporting SSD which stores the atomicity information in the spare area of the flash memory page. We evaluate the performance of the proposed atomic write scheme in MariaDB using the tpcc-mysql and SysBench benchmarks. The experimental results show that the proposed atomic write technique shows a performance improvement of 1.4~1.5 times compared to the double write buffer technique.

• Journal of the Korea Society of Computer and Information
• /
• v.19 no.7
• /
• pp.113-130
• /
• 2014

In this paper, we propose BLOCS algorithm to find sequence of data that should be saved in cache device of hybrid storage system which uses SSD as a cache device. BLOCS algorithm which uses a sequence pattern mining scheme, creates a set of frequently requested sectors with respect to requested order of sectors. To compare the performance of the proposed scheme, we introduce Distance (DIST) based scheme, Request Frequency (FREQ) based scheme, and Frequency times Size (F-S) based scheme. We measure the hit ratio and I/O latency of different caching schemes using hybrid storage caching simulator. We acquired booting workload along with ten scenarios of launching applications and use the workloads as input to the cache simulator. After experiment with booting workload, we find that BLOCS scheme gives hit ratio of 61% which is about 15% higher than the least performing DIST scheme.

• Proceedings of the KAIS Fall Conference
• /
• 2010.05a
• /
• pp.176-178
• /
• 2010

본 논문에서는 전통적인 RAID-1 미러링에 기반을 둔 헤테로-미러링이라는 하는 새로운 저장 장치 관리 기법을 제안한다. 헤테로-미러링 기반의 스토리지 관리는 SSD에서 발생 가능한 프리징 현상을 피하기 위한 쓰기-부하 밸런싱과 쓰기 지연 연산을 통하여 RAID-1 처리 성능을 개선한다.