• Annual Conference of KIPS
• /
• 2014.11a
• /
• pp.99-101
• /
• 2014

다양한 응용 분야에서 대량의 데이타 저장이 빈번해 짐에 따라 분산 파일 시스템에 대한 효용성이 커지고 있으며, 이로 인해 수많은 분산 파일 시스템이 개발되고 있다. 이러한, 분산 파일 시스템의 효율성을 검증할 수 있는 방법에 대한 필요성이 커지고 있다. 또한, SSD와 같은 우수한 성능을 가지는 장치를 사용하는 분산 파일 시스템에서는 SSD와 같은 장치의 효율성을 보기 위한 복잡한 워크로드를 생성 하여 시험할 수 있어야 한다. 하지만 기존의 입출력 시험 도구에서는 복잡한 상황을 시험하는 것이 매우 어렵다. 따라서, 본 논문에서는 필요에 따라 파일의 분포에 따라 다른 입출력 패턴을 가지는 여러 부하 상황을 생성이 가능하며, 웹을 통해 손쉽게 시험할 수 있는 분산 파일 시스템을 위한 웹기반 워크로드 생성기를 개발하였다.

• KIISE Transactions on Computing Practices
• /
• v.23 no.6
• /
• pp.379-385
• /
• 2017

In enterprise environments, hybrid storage typically utilizes SSDs over disk-based RAID. Typically, SSDs over RAID are used as the data cache. Recently, the LeavO caching scheme was introduced to reduce the parity update overhead of the underlying RAID. In this paper, we combine the data caching and LeavO caching schemes and derive cost models of the combined cache to determine the optimal data and LeavO cache sizes. We also propose the Adaptive Combined Cache that dynamically adjusts the data cache and LeavO cache sizes for evolving workloads. Experimental results show that the performance of the Adaptive Combined Cache is significantly superior to that of the conventional data caching scheme and is comparable with that of the off-line optimal scheme.

• Annual Conference of KIPS
• /
• 2015.04a
• /
• pp.676-679
• /
• 2015

빅데이터 분석을 위한 Hadoop 환경에서 고성능 저장장치인 SSD에 대한 중요성이 증가하면서 일반적으로 사용되는 저장장치인 HDD와 혼합하여 사용하는 연구들이 주목 받고 있다. 특히 SSD를 HDD의 캐시로 사용했을 때 저장장치에 대한 I/O 성능을 향상할 수 있다는 연구 결과들이 있다. 본 연구는 이를 바탕으로 SSD를 HDD의 캐시로 사용한다. HDFS는 저장장치에 접근하여 I/O를 수행하는데 기존에는 로컬 서버에서 캐시 미스가 발생한 경우 로컬 HDD로 접근한다. 이러한 방식은 접근하는 데이터에 따라 SSD의 높은 Bandwidth를 활용하지 못하게 되는 경우를 발생시키고 그 결과 특정 서버의 I/O 지연으로 전체 분산 처리의 성능을 저하시킬 수 있다. 이를 해결하기 위해 본 연구는 HDFS 레벨에서 로컬 서버의 HDD와 데이터 복제본들이 저장된 원격 서버의 SSD에서 I/O를 수행하는 경우에 대해 수식을 통해 비용을 비교한다. 그 결과 항상 기대 성능이 높은 저장 장치를 선택하여 데이터를 읽어오게 함으로써 기존 방식보다 성능이 개선될 수 있음을 입증한다.

• Journal of KIISE
• /
• v.44 no.8
• /
• pp.767-773
• /
• 2017

In a distributed system environment, such as a cloud database, the tail latency needs to be kept short to ensure uniform quality of service. In this paper, through experiments on a Cassandra database, we show that long tail latency is caused by a lack of memory space because the database cannot receive any request until free space is reclaimed by writing the buffered data to the storage device. We observed that, since the performance of the storage device determines the amount of time required for writing the buffered data, the performance degradation of Solid State Drive (SSD) due to garbage collection results in a longer tail latency. We propose a garbage collection synchronization technique, called SyncGC, that simultaneously performs garbage collection in the java virtual machine and in the garbage collection in SSD concurrently, thus hiding garbage collection overheads in the SSD. Our evaluations on real SSDs show that SyncGC reduces the tail latency of $99.9^{th}$ and, $99.9^{th}-percentile$ by 31% and 36%, respectively.

• Journal of Digital Contents Society
• /
• v.14 no.1
• /
• pp.15-23
• /
• 2013

In response to the demands of large-scale data processing with low-power and new application, a storage system using SSD (Solid State Disk/Drive) with fast input-output performance instead of hard disc has appeared as storage device. Studies on methods to overcome specific problems of SSD such as various processing data units, out-place-update and limited delete count have been actively conducted. However, declining performance and stability have not been resolved yet when storing case specific data with small scale that causes frequent random write in hard disc or SSD. This thesis suggests a system structure that stores index requesting frequent random write in NVRAM capable of byte access by using characteristics such as byte unit fast read / write of NVRAM, non-volatile and smaller size of actual changed data size in index page than block size.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.48 no.8
• /
• pp.45-52
• /
• 2011

Flash memory is becoming widely prevalent in various area due to high performance, non-volatile features, low power, and robust durability. As price-per-bit is decreased, NAND flash based SSDs (Solid State Disk) have been attracting attention as the next generation storage device, which can replace HDDs (Hard Disk Drive) which have mechanical properties. Especially for the single package SSD, if channel number or FIFO buffer size per channel increases to improve performance, the size of a controller and I/O pin count will increase linearly with channel numbers and form factor will be affected. We propose a novel technique which can minimize form factor by optimizing the number of NAND flash channels and the size of interface FIFO buffer in the SSD. For SSD with 10 channel and double buffer, the experimental results show that buffer block size can be reduced about 73% without performance degradation and total size of a controller can be reduced about 40% because control block per channel and I/O pin count decrease according to decrease channel number.

• Journal of Digital Contents Society
• /
• v.13 no.1
• /
• pp.75-81
• /
• 2012

SSDs(Solid State Drives) have many attractive features such as high performance, low power consumption, shock resistance, and low weight, so they replace HDDs to a certain extent. An SSD has FTL(Flash Translation Layer) which emulate block storage devices like HDDs. A garbage collection, one of major functions of FTL, effects highly on the performance and the lifetime of SSDs. However, there is no de facto standard for new garbage collection algorithms. To solve this problem, we propose trace driven offline optimal algorithms for garbage collection of FTL. The proposed algorithm always guarantees minimal number of erase operation. In addition, we verify our proposed algorithm using TPC trace.

• Journal of KIISE
• /
• v.43 no.2
• /
• pp.177-185
• /
• 2016

In this paper, we propose a hybrid write buffer architecture comprised of DRAM and NVRAM on SSD and a write buffer algorithm for the hybrid write buffer architecture. Unlike other write buffer algorithms, the proposed algorithm considers read pages as well as write pages to improve the performance of storage devices because most actual workloads are read-write mixed workloads. Through effectively managing NVRAM pages, the proposed algorithm extends the endurance of SSD by reducing the number of erase operations on NAND flash memory. Our experimental results show that our algorithm improved the buffer hit ratio by up to 116.51% and reduced the number of erase operations of NAND flash memory by up to 56.66%.

• Journal of KIISE:Computing Practices and Letters
• /
• v.15 no.9
• /
• pp.685-689
• /
• 2009

Flash memory has several features such as low~power consumption and fast access so that there has been various research on using flash memory as new storage. Especially the Solid State Disk which is composed of flash memory chips has recently replaced the hard disk. At present, SSD adopts the multi-channel and multi-way architecture to exploit advantages of parallel access. In this architecture, data are written on SSD in a unit of a superblock which is composed of multiple blocks in which some blocks are put together. This paper proposes two schemes of selecting, segmenting and re-composing victim superblocks to optimize concurrent processing when a buffer flush occurs. The experimental results show that 35% of superblock- based write operations is reduced by selecting victims and additional 9% by composition of superblock.

• Journal of Internet Computing and Services
• /
• v.16 no.5
• /
• pp.19-27
• /
• 2015

Due to the advantages of fast read/write operation and low power consumption, SSD(Solid State Drive) is now widely adopted as storage device of smart phone, laptop computer, server, etc. However, the shortcomings of SSD such as limited number of write operations and asymmetric read/write operation lead to the problem of shortened life span of SSD. Therefore, the block replacement policy of SSD used as cache for HDD is very important. The existing solutions for improving the lifespan of SSD including the LARC scheme typically employ the LRU algorithm to manage the SSD blocks, which may increase the miss rate in SSD due to the replacement of frequently used block instead of rarely used block. In this paper we propose a novel block replacement scheme which considers the block reuse interval to effectively handle various data read/write patterns. The proposed scheme replaces the block in SSD based on the recency decided by reuse interval and age along with hit ratio. Computer simulation using workload trace files reveals that the proposed scheme consistently improves the performance and lifespan of SSD by increasing the hit ratio and decreasing the number of write operations compared to the existing schemes including LARC.