Results 1 -
8 of
8
Comparing VM-placement algorithms for on-demand Clouds
- IN PROCEEDINGS OF IEEE INTERNATIONAL CONFERENCE ONCLOUD COMPUTING TECHNOLOGY AND SCIENCE. LOS ALAMITOS: IEEE COMPUTER SOCIETY
, 2011
"... Much recent research has been devoted to investigating algorithms for allocating virtual machines (VMs) to physical machines (PMs) in infrastructure clouds. Many such algorithms address distinct problems, such as initial placement, consolidation, or tradeoffs between honoring service-level agreement ..."
Abstract
-
Cited by 29 (4 self)
- Add to MetaCart
(Show Context)
Much recent research has been devoted to investigating algorithms for allocating virtual machines (VMs) to physical machines (PMs) in infrastructure clouds. Many such algorithms address distinct problems, such as initial placement, consolidation, or tradeoffs between honoring service-level agreements and constraining provider operating costs. Even where similar problems are addressed, each individual research team evaluates proposed algorithms under distinct conditions, using various techniques, often targeted to a small collection of VMs and PMs. In this paper, we describe an objective method that can be used to compare VMplacement algorithms in large clouds, covering tens of thousands of PMs and hundreds of thousands of VMs. We demonstrate our method by comparing 18 algorithms for initial VM placement in on-demand infrastructure clouds. We compare algorithms inspired by open-source code for infrastructure clouds, and by the online bin-packing literature.
VM Leakage and Orphan Control in Open-Source Clouds
"... Abstract—Computer systems often exhibit degraded performance due to resource leakage caused by erroneous programming or malicious attacks, and computers can even crash in extreme cases of resource exhaustion. The advent of cloud computing provides increased opportunities to amplify such vulnerabilit ..."
Abstract
-
Cited by 3 (2 self)
- Add to MetaCart
(Show Context)
Abstract—Computer systems often exhibit degraded performance due to resource leakage caused by erroneous programming or malicious attacks, and computers can even crash in extreme cases of resource exhaustion. The advent of cloud computing provides increased opportunities to amplify such vulnerabilities, thus affecting a significant number of computer users. Using simulation, we demonstrate that cloud computing systems based on open-source code could be subjected to a simple malicious attack capable of degrading availability of virtual machines (VMs). We describe how the attack leads to VM leakage, causing orphaned VMs to accumulate over time, reducing the pool of resources available to users. We identify a set of orphan control processes needed in multiple cloud components, and we illustrate how such processes detect and eliminate orphaned VMs. We show that adding orphan control allows an open-source cloud to sustain a higher level of VM availability during malicious attacks. We also report on the overhead of implementing orphan control. Keywords- availability; cloud computing; modeling; reliability; scalable fault resilience techniques I.
Date: Approved:
, 2012
"... Cloud computing appears to be a paradigm shift in service oriented computing. Massively scalable Cloud architectures are spawned by new business and social ap-plications as well as Internet driven economics. Besides being inherently large scale and highly distributed, Cloud systems are almost always ..."
Abstract
- Add to MetaCart
(Show Context)
Cloud computing appears to be a paradigm shift in service oriented computing. Massively scalable Cloud architectures are spawned by new business and social ap-plications as well as Internet driven economics. Besides being inherently large scale and highly distributed, Cloud systems are almost always virtualized and operate in automated shared environments. The deployed Cloud services are still in their infancy and a variety of research challenges need to be addressed to predict their long-term behavior. Performance and dependability of Cloud services are in general stochastic in nature and they are affected by a large number of factors, e.g., nature of workload and faultload, infrastructure characteristics and management policies. As a result, developing scalable and predictive analytics for Cloud becomes difficult and non-trivial. This dissertation presents the research framework needed to de-velop high fidelity stochastic models for large scale enterprise systems using Cloud computing as an example. Throughout the dissertation, we show how the developed models are used for: (i) performance and availability analysis, (ii) understanding
Practical Challenges when Implementing a Distributed Population of Cloud-Computing Simulators Controlled by a Genetic Algorithm
"... Abstract—The recent explosion of affordable multicore, multichip systems, coupled with cluster management software, encourages the development of novel distributed applications for exploring large parameter spaces. We expect many such applications will soon appear. For example, we recently applied a ..."
Abstract
- Add to MetaCart
Abstract—The recent explosion of affordable multicore, multichip systems, coupled with cluster management software, encourages the development of novel distributed applications for exploring large parameter spaces. We expect many such applications will soon appear. For example, we recently applied a genetic algorithm to steer a population of cloud-computing simulators toward low-probability, costly failure scenarios. We aim to provide a design-time tool that system engineers can use to identify and mitigate such scenarios. We found that our idea was much simpler in theory than in practice, largely due to implementation challenges that arose. In this paper, we describe the design and deployment of our application, and we identify and discuss the practical challenges that we faced. We outline pragmatic solutions that we adopted to overcome those challenges. We believe many near-future applications will face similar challenges, so we hope that our experiences prove instructive. Index Terms — Computational steering, cloud computing, cluster computing, discrete event simulation, distributed systems, fault tolerance, genetic algorithms, software for parallel and distributed systems I.
HARDWARE-IN-THE-LOOP SIMULATION FOR AUTOMATED BENCHMARKING OF CLOUD INFRASTRUCTURES
"... To address the challenge of automated performance benchmarking in virtualized cloud infrastructures, an extensible and adaptable framework called CloudBench has been developed to conduct scalable, controllable, and repeatable experiments in such environments. This paper presents the hardware-in-the- ..."
Abstract
- Add to MetaCart
To address the challenge of automated performance benchmarking in virtualized cloud infrastructures, an extensible and adaptable framework called CloudBench has been developed to conduct scalable, controllable, and repeatable experiments in such environments. This paper presents the hardware-in-the-loop simulation technique used in CloudBench, which integrates an efficient discrete-event simulation with the cloud infrastructure under test in a closed feedback control loop. The technique supports the decomposition of complex resource usage patterns and provides a mechanism for statistically multiplexing application requests of varied characteristics to generate realistic and emergent behavior. It also exploits parallelism at multiple levels to improve simulation efficiency, while maintaining temporal and causal relationships with proper synchronization. Our experiments demonstrate that the proposed technique can synthesize complex resource usage behavior for effective cloud performance benchmarking. 1
Combining Genetic Algorithms and Simulation to Search for Failure Scenarios in System Models
"... Abstract—Large infrastructures, such as clouds, can exhibit substantial outages, sometimes caused by failure scenarios not predicted during system design. We define a method for model-based prediction of system quality characteristics. The method uses a genetic algorithm to search system simulations ..."
Abstract
- Add to MetaCart
Abstract—Large infrastructures, such as clouds, can exhibit substantial outages, sometimes caused by failure scenarios not predicted during system design. We define a method for model-based prediction of system quality characteristics. The method uses a genetic algorithm to search system simulations for parameter combinations that result in system failures, so that designers can take mitigation steps before deployment. We apply the method to study an existing infrastructure-as-a-service cloud simulator. We characterize the dynamics, quality, effectiveness and cost of genetic search, when applied to seek a known failure scenario. Further, we iterate the search to reveal previously unknown failure scenarios. We find that, when schedule permits and failure costs are high, combining genetic search with simulation proves useful for exploring and improving system designs. Keywords—genetic algorithms; model-based prediction; simulation methodology; system design I.
