• 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.

• Journal of applied mathematics & informatics
• /
• v.8 no.3
• /
• pp.1035-1045
• /
• 2001

We present two decomposition approximations for the mean sojourn times in open single class queing networks. If there is a single bottleneck station, the approximations are asymptotically exact in both light and heavy traffic. When applied to a Jackson network or an M/G/1 queue, these approximations are exact for all values of the traffic intensity.

• Proceedings of the Korean Information Science Society Conference
• /
• 2001.04a
• /
• pp.316-318
• /
• 2001

통합 서비스 망은 다양한 트래픽 소스의 다양한 지원을 필요로 한다. 본 논문은 패킷 스케줄링 알고리즘 중 WFQ와 SFQ를 비교 분석하여 성능을 비교한 후 다양한 트래픽 소스를 전송하기 위해 제안된 방법을 WFQ와 SFQ 스케줄링 알고리즘에 적용하여 비교 분석 할 것이다.

• The Journal of the Korea Contents Association
• /
• v.8 no.1
• /
• pp.300-307
• /
• 2008

Power management (PM) depends on the power state transition and system workload. The system model is composed of corresponding stochastic models of the power state and system queue. In this paper, stochastic models which can handle various PM techniques are developed. SRN (Stochastic Reward Nets), an extended Petri-Net, has facilities that represent system queue and various modelling functions. The SRN is employed for developing PM models. An adaptive timeout PM model is also introduced and the power consumption and performance of this model are compared with other existing PM techniques models such as greedy and N-Policy techniques.

• Journal of applied mathematics & informatics
• /
• v.17 no.1_2_3
• /
• pp.737-746
• /
• 2005

Average memory occupancy and congestion in computer system or communication system may be reduced further if new jobs are admitted only when the number of jobs queued at CPU is below a certain threshold, run queue cutoff (RQ). In our previous paper we showed that response time of a job is invariant with respect to RQ if jobs do not communicate each other. In this paper, we prove that the invariance property by considering the evolution of the queue lengths as point processes. We also present an approximate method for the delay due to context switching under time slicing.

• Journal of applied mathematics & informatics
• /
• v.17 no.1_2_3
• /
• pp.547-557
• /
• 2005

In this paper, we investigate the packet delay distribution of a dynamic bandwidth allocation(DBA) scheme in an Ethernet passive optical network(EPON). We focus on the gated-exhaustive vacation system. We assume that input packets arrive at an optical network unit(ONU) according to general interarrival distribution. We use a discrete time queueing model in order to find the packet delay distribution of the gated-exhaustive system with the primary transmission queue and the secondary input queue. We give some numerical examples to investigate the mean packet delays of the proposed queueing model to analyze the DBA scheme in an EPON.

• Proceedings of the Korean Information Science Society Conference
• /
• 2012.06a
• /
• pp.313-315
• /
• 2012

SSD(Solid State Drives)에서는 쓰기 버퍼를 활용해 쓰기 연산의 횟수를 줄임으로써 입출력 성능향상을 가져오고, 부가적으로 삭제 연산이 줄어들어 수명 향상의 효과를 얻고 있다. 하지만 지금까지의 쓰기 버퍼 관련 연구에서는 SSD의 컨트롤러 구조를 모두 고려하지 못하고 있다. 이 논문에서는 쓰기 버퍼의 데이터 히트(HIT)율을 높이기 위한 방법으로 지금까지 고려하지 못했던 SSD 컨트롤러 구조 내 명령 큐(Command Queue)의 정보를 통해 가까운 미래에 도착할 쓰기 명령을 예측하고, 이를 기존 버퍼 관리 기법에 적용하는 방안을 연구한다.

• 제어로봇시스템학회:학술대회논문집
• /
• 2004.08a
• /
• pp.767-771
• /
• 2004

In this paper the performance of some control theory-based Active Queue Management (AQM) methods are improved and investigated through simulation. The simulation scenario is considered to be closely realistic, in which the traffic is mainly due to http sessions. The results show the best performance of PID controller with respect to the others in good queue regulation as well as high link utilization and low delay metrics.

• Journal of applied mathematics & informatics
• /
• v.7 no.3
• /
• pp.861-873
• /
• 2000

Rate-based flow control plays an important role for efficient traffic management of Available Bit Rate(ABR) services. We deal with the problem of the buffer dimension for rate-based ABR control. In this paper, we analyze the Allowed Cell Rate(ACR) of a source and the queue size in a steady state. First, we investigate the effect of the number of active sources on the behavior of the ACR and the maximum queue size. Reflecting the effect of this real scenes, we determine the optimal buffer size and buffer threshold. Furthermore, our analytic results are compared with the case when the effect of the number of active sources is disregarded.

• Journal of applied mathematics & informatics
• /
• v.16 no.1_2
• /
• pp.539-550
• /
• 2004

We consider communication or computer systems that have the average memory occupancy and congestion. A certain system may be reduced further if new processes are admitted only when the number of processes queued at the processor is below a certain threshold, run queue cutoff(RQ). We show that response time of a process is invariant with respect to RQ when processes do not communicate each other. The invariance property is considered with the evolution of the queue lengths as point processes and demonstrate it numerically using matrix-geometric methods. We illustrate the RQ invariance property.