In virtualized environments, physical resources are abstracted as resource pool for provisioning, services consolidation, and service on demand. Traditional virtual machines are often over-provisioned to provide peak performance guarantees and thus waste a lot of memory resources. In this paper we propose a novel Initiatively Share Based Memory Overcommitting scheme, namely, IMR, to improve the memory utilization further. In IMR, guest virtual machine kernels are hacked with initiatively share functionality such that the guests can cooperatively share unused memory with the hypervisor, and consequently some of the possible page swapping and merging in such virtual machines are eliminated. The experimental results in shadow page tables and extended page tables scenarios show that our IMR approach not only improves the memory utilization but also has microsecond-level time overheads for shared memory reclaim, which outperforms the memory optimization schemes such as millisecond-level KSM (Kernel Same-page Merging) in best cases and millisecond-level page swapping at average.
목차
Abstract 1. Introduction 2. Related Works 2.1. Memory Management in KVM 2.2. Memory Management Status on Other Platforms 2.3. Analysis on Memory Virtualization 3. IMR Design 3.1. Design Principle and Framework 3.2. IMR Performance Modeling 4. Experiments and Analysis 4.1. Experiment Environment 4.2. Experiment Scheme and Analysis 5. Conclusions Acknowledgements References
보안공학연구지원센터(IJGDC) [Science & Engineering Research Support Center, Republic of Korea(IJGDC)]
설립연도
2006
분야
공학>컴퓨터학
소개
1. 보안공학에 대한 각종 조사 및 연구
2. 보안공학에 대한 응용기술 연구 및 발표
3. 보안공학에 관한 각종 학술 발표회 및 전시회 개최
4. 보안공학 기술의 상호 협조 및 정보교환
5. 보안공학에 관한 표준화 사업 및 규격의 제정
6. 보안공학에 관한 산학연 협동의 증진
7. 국제적 학술 교류 및 기술 협력
8. 보안공학에 관한 논문지 발간
9. 기타 본 회 목적 달성에 필요한 사업
간행물
간행물명
International Journal of Grid and Distributed Computing
간기
격월간
pISSN
2005-4262
수록기간
2008~2016
십진분류
KDC 505DDC 605
이 권호 내 다른 논문 / International Journal of Grid and Distributed Computing Vol.7 No.4