RISS 학술연구정보서비스

검색

인기 검색어

    다국어 입력

    http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.

    변환된 중국어를 복사하여 사용하시면 됩니다.

    예시)
    • 中文 을 입력하시려면 zhongwen을 입력하시고 space를누르시면됩니다.
    • 北京 을 입력하시려면 beijing을 입력하시고 space를 누르시면 됩니다.
    닫기

    쓰기 집중 환경에서의 접근 빈도와 병합 효율을 반영한 SSD 저장 장치 기반 하이브리드 디스크 버퍼 정책

    한글로보기

    https://www.riss.kr/link?id=T17279345

    • 0

      상세조회
    • 0

      다운로드
    서지정보 열기
    • 내보내기
    • 내책장담기
    • 공유하기
    • 오류접수

    부가정보

    다국어 초록 (Multilingual Abstract) kakao i 다국어 번역

    Hybrid storage systems that integrate DRAM and non-volatile memory (NVM) have gained attention as an effective solution to enhance the endurance and performance of solid-state drives (SSD). However, traditional buffer replacement policies tend to focus solely on access frequency, without considering the sequential characteristics of write operations. This limitation leads to inefficient merge behavior in the Flash Translation Layer (FTL) and results in increased write amplification.
    To address this issue, this work presents a hybrid buffer management architecture named CLOCK-LOG-DPP, which integrates two complementary policies. CLOCK-DPP is a frequency-aware policy that retains dirty pages in DRAM to reduce write pressure on NVM, while LOG-DPP is a merge-aware strategy that clusters logically adjacent pages and prioritizes those with high sequentiality scores for eviction to promote switch merges.
    CLOCK-DPP reduces the total write count, block erasures, and energy consumption by up to 23.1%, 19.8%, and 25.7%, respectively, compared to baseline methods. LOG-DPP further enhances write efficiency, reducing write count by up to 38.6%, block erasures by 34.2%, and energy usage by 41.3%, while maintaining comparable cache hit ratios.
    When deployed as a unified framework, CLOCK-LOG-DPP achieves up to 40.9% reduction in overall write amplification, while significantly improving SSD endurance and energy efficiency. These results demonstrate the effectiveness of combining frequency-aware and merge-aware strategies in hybrid SSD buffer management.
    번역하기

    Hybrid storage systems that integrate DRAM and non-volatile memory (NVM) have gained attention as an effective solution to enhance the endurance and performance of solid-state drives (SSD). However, traditional buffer replacement policies tend to focu...

    Hybrid storage systems that integrate DRAM and non-volatile memory (NVM) have gained attention as an effective solution to enhance the endurance and performance of solid-state drives (SSD). However, traditional buffer replacement policies tend to focus solely on access frequency, without considering the sequential characteristics of write operations. This limitation leads to inefficient merge behavior in the Flash Translation Layer (FTL) and results in increased write amplification.
    To address this issue, this work presents a hybrid buffer management architecture named CLOCK-LOG-DPP, which integrates two complementary policies. CLOCK-DPP is a frequency-aware policy that retains dirty pages in DRAM to reduce write pressure on NVM, while LOG-DPP is a merge-aware strategy that clusters logically adjacent pages and prioritizes those with high sequentiality scores for eviction to promote switch merges.
    CLOCK-DPP reduces the total write count, block erasures, and energy consumption by up to 23.1%, 19.8%, and 25.7%, respectively, compared to baseline methods. LOG-DPP further enhances write efficiency, reducing write count by up to 38.6%, block erasures by 34.2%, and energy usage by 41.3%, while maintaining comparable cache hit ratios.
    When deployed as a unified framework, CLOCK-LOG-DPP achieves up to 40.9% reduction in overall write amplification, while significantly improving SSD endurance and energy efficiency. These results demonstrate the effectiveness of combining frequency-aware and merge-aware strategies in hybrid SSD buffer management.

    더보기

    목차 (Table of Contents)

    • 1. 서론 1
    • 1.1 연구 배경 1
    • 1.2 연구 목적 3
    • 1.3 논문 구성 6
    • 2. 배경지식 및 관련 연구 7
    • 1. 서론 1
    • 1.1 연구 배경 1
    • 1.2 연구 목적 3
    • 1.3 논문 구성 6
    • 2. 배경지식 및 관련 연구 7
    • 2.1 SSD 7
    • 2.2 디스크 버퍼 11
    • 2.3 플래시 변환 계층 15
    • 2.4 CLOCK 알고리즘 23
    • 2.5 블록 병합 기법 26
    • 3. CLOCK-DPP: SSD의 접근 빈도와 쓰기 상태 기반의 하이브리드 디스크 버퍼 관리 정책 29
    • 3.1 하이브리드 디스크 버퍼 31
    • 3.2 접근 빈도와 쓰기 여부를 고려한 하이브리드 디스크 버퍼 정책 36
    • 3.3 시나리오 기반의 페이지 처리 흐름 44
    • 4. LOG-DPP: SSD에서 순차성과 병합 가능성을 고려한 하이브리드 디스크 버퍼 교체 정책 47
    • 4.1 Write Buffer Cache 환경에서의 디스크 버퍼 49
    • 4.2 순차성과 병합 가능성을 고려한 클러스터 기반 하이브리드 디스크 버퍼 정책 55
    • 4.3 시나리오 기반의 클러스터 처리 흐름 61
    • 5. 실험 및 성능 평가 65
    • 5.1 실험 환경 65
    • 5.2 실험 결과 67
    • 5.2.1 디스크 버퍼 총 쓰기 횟수 67
    • 5.2.2 에너지 소모량 71
    • 5.2.3 디스크 버퍼 적중률 74
    • 5.2.4 블록 소거 횟수 77
    • 6. 결론 80
    • 참고문헌 82
    • Abstract 89
    더보기

    분석정보

    View

    상세정보조회

    0

    Usage

    원문다운로드

    0

    대출신청

    0

    복사신청

    0

    EDDS신청

    0

    동일 주제 내 활용도 TOP

    더보기

    주제

    연도별 연구동향

    연도별 활용동향

    연관논문

    연구자 네트워크맵

    공동연구자 (7)

    유사연구자 (20) 활용도상위20명

    이 자료와 함께 이용한 RISS 자료

    나만을 위한 추천자료

    해외이동버튼