CAJ | 학술논문

针对云计算环境中资源过度供给的问题,提出一种基于弹性预留机制的虚拟机供给策略.该策略在按需供给的基础上引入弹性预留机制,通过分析用户预留请求的统计特征,允许系统在一定条件下接纳用户的超额预留请求,从而在保证用户满意度的前提下提高云系统的资源有效利用率.理论分析给出了弹性预留机制的概率计算方法,实验分析则验证了所提资源供给策略的有效性以及关键策略参数对系统性能的影响.实验结果显示,与已有虚拟资源供给策略相比,本文所提策略能够有效提高云系统中的资源有效利用率,同时降低过度预留所导致的短期资源供给不足问题.
침대운계산배경중자원과도공급적문제,제출일충기우탄성예류궤제적허의궤공급책략.해책략재안수공급적기출상인입탄성예류궤제,통과분석용호예류청구적통계특정,윤허계통재일정조건하접납용호적초액예류청구,종이재보증용호만의도적전제하제고운계통적자원유효이용솔.이론분석급출료탄성예류궤제적개솔계산방법,실험분석칙험증료소제자원공급책략적유효성이급관건책략삼수대계통성능적영향.실험결과현시,여이유허의자원공급책략상비,본문소제책략능구유효제고운계통중적자원유효이용솔,동시강저과도예류소도치적단기자원공급불족문제.
To address the problem of resource overprovision in cloud environments,a virtual machine provision policy based on elastic reservation mechanism is proposed.It extends the conventional ＂ondemand provision＂ by introducing a novel elastic reservation mechanism,which allows cloud providers to accept user’s over-reservation requests under certain conditions so as to improve the resource effective utilization and user’s satisfaction.Theoretical analysis presents the approach to calculating the probability of elastic reservation,and experimental evaluation validates the effectiveness of the proposed policy and the effects of key parameter on system performance.The results show that comparing with existing techniques,elastic reservation based resource provision policy can significantly increase the resource effective utilization,and decrease the short-term resource under-provision caused by excessively reservations.