Reliability Aware Cost Optimization for Memory Constrained Cloud Workflows

Infrastructure as a service (IaaS) allows users to rent resources from the Cloud to meet their various computing requirements. The pay-as-you-use model, however, poses a nontrivial technical challenge to the IaaS cloud service providers: how to fast provision a large number of virtual machines (VMs) to meet users' dynamic computing requests? We address this challenge with VMThunder, a new VM provisioning tool, which downloads data blockson demandduring the VM booting process and speeds up VM image streaming by strategically integrating peer-to-peer (P2P) streaming techniques with enhanced optimization schemes such as transfer on demand, cache on read, snapshot on local, and relay on cache. In particular, VMThunder stores the original images in a share storage and in the meantime it adopts a tree-based P2P streaming scheme so that common image blocks are cached and reused across the nodes in the cluster. We implement VMThunder in CentOS Linux and thoroughly test its performance. Comprehensive experimental results show that VMThunder outperforms the state-of-the-art VM provisioning methods, with respect to scalability, latency, and VM runtime I/O performance.

A large number of virtual machines (VMs) have been developed for sensor nodes. Many are based on Java, although implementations based on Python and.Net also exist. Some aspects of the java virtual machine, such as the simple yet sufficiently expressive instruction set, work well on a sensor node. However, Java was developed for more powerful devices, which results in certain functionality not being ideal for resource constrained devices. While writing VM implementations and programmes to run on sensor node devices over the past decade, we encountered a number of situations where this mismatch between standard languages and the specifics of sensor node programming limited the practical use of current VMs. In this letter, we describe what we consider to be the most pressing issues and discuss how these may be resolved in future VMs.

Cloud computing offers new computing paradigms, capacity and flexible solutions to high performance computing (HPC) applications. For example, Hardware as a Service (HaaS) allows users to provide a large number of virtual machines (VMs) for computation-intensive applications using the HaaS model. Due to the large number of VMs and electronic components in HPC system in the cloud, any fault during the execution would result in re-running the applications, which will cost time, money and energy. In this paper we presented a proactive fault tolerance (FT) approach to HPC systems in the cloud to reduce the wall-clock execution time and dollar cost in the presence of faults. We also developed a generic FT algorithm for HPC systems in the cloud. Our algorithm does not rely on a spare node prior to prediction of a failure. We also developed a cost model for executing computation-intensive applications on HPC systems in the cloud. We analysed the dollar cost of provisioning spare nodes and checkpointing FT to assess the value of our approach. Our experimental results obtained from a real cloud execution environment show that the wall-clock execution time and cost of running computation-intensive applications in cloud can be reduced by as much as 30%. The frequency of checkpointing of computation-intensive applications can be reduced up to 50% with our FT approach for HPC in the cloud compared with current FT approaches.

The virtual machine (VM) scheduling problem in cloud brokers that support cloud bursting is fraught with uncertainty due to the on-demand nature of Infrastructure as a Service (IaaS) VMs. Until a VM request is received, the scheduler does not know in advance when it will arrive or what configurations it demands. Even when a VM request is received, the scheduler does not know when the VM's lifecycle expires. Existing studies begin to use deep reinforcement learning (DRL) to solve such scheduling problems. However, they do not address how to guarantee the QoS of user requests. In this paper, we investigate a cost optimization problem for online VM scheduling in cloud brokers for cloud bursting to minimize the cost spent on public clouds while satisfying specified QoS restrictions. We propose DeepBS, a DRL-based online VM scheduler in a cloud broker which learns from experience to adaptively improve scheduling strategies in environments with non-smooth and uncertain user requests. We evaluate the performance of DeepBS under two request arrival patterns which are respectively based on Google and Alibaba cluster traces, and the experiments show that DeepBS has a significant advantage over other benchmark algorithms in terms of cost optimization.

Energy-efficient VM opening algorithms for real-time workflows in heterogeneous clouds

The public cloud environment has emerged as a promising platform for exe-cuting scientific workflows. These executions involve leasing virtual machines (VMs) from public services for the duration of the workflow. The structure of the workflows significantly impacts the performance of any proposed scheduling approach. A task within a workflow cannot begin its execution before receiving all the required data from its preceding tasks. In this paper, we introduce a multi-priority scheduling approach for executing workflow tasks in the cloud. The key component of the proposed approach is a mechanism that logically or-ders and groups workflow tasks based on their data dependencies and locality. Using the proposed approach, the number of available VMs influences the num-ber of groups (partitions) obtained. Based on the locality of each group’s tasks, the priority of each group is determined to reduce the overall execution delay and improve VM utilization. As the results demonstrate, the proposed approach achieves a significant reduction in both execution costs and time in most scenar-ios

Recent years witness the proliferation of Infrastructure-as-a-Service (IaaS) cloud services, which provide on-demand resources (CPU, RAM, disk) in the form of virtual machines (VMs) for hosting applications/services of third parties. Given the state-of-the-art IaaS offerings, it is still a problem of fundamental importance how the Application Service Providers (ASPs) should rent VMs from the clouds to serve their application needs, in order to minimize the cost while meeting their job demands over a long run. Cloud providers offer different pricing options to meet computing requirements of a variety of applications. However, the challenge facing an ASP is how these pricing options can be dynamically combined to serve arbitrary demands at the optimal cost. In this paper, we propose an online VM purchasing algorithm based on the Lyapunov optimization technique, for minimizing the long-term-averaged VM rental cost of an ASP with time-varying and delay-tolerant workloads, while bounding the maximum response delay of its jobs. In stark contrast with the existing studies, the proposed algorithm enables an ASP to optimally decide the amount of reserved, on-demand and spot instances to purchase simultaneously. Rigorous analysis shows that our algorithm can achieve a time-averaged resource cost close to the offline optimum. Trace-driven simulations further verify the efficacy of our algorithm.

In recent years, intelligent flow sensors have been applied to many fields. Cloud operation is a design method to further improve the intelligence of such sensors. However, the cloud workflows of intelligent flow sensors consume too much energy, making it imperative to schedule cloud workflows. With the growing awareness of energy conservation, it is a hot topic to design an energy-efficient workflow scheduling algorithm. Therefore, this paper puts forward the predict minimum energy consumption (PMEC) algorithm, a cloud workflow scheduling algorithm that strikes a balance between energy consumption and execution time. Firstly, the optimistic cost table (OCT) was adopted to rank the tasks by priority. Then, the resources, i.e. virtual machines, were assigned statically to the tasks, in the light of task priority and energy consumption. After that, the workflow was scheduled according to the assignments. Simulation results show that the PMEC is much more energy efficient than traditional listbased scheduling algorithms.

To address the virtual machine cluster deployment issues in cloud computing environment, a novel MCSA (Min-cut segmentation algorithm) of virtual machine cluster is proposed with resource and communication bandwidth constraints. In this paper, the basic idea is based on the fully consideration on the CPU, memory, hard-disk and other resource constraints between virtual machine cluster and physical host, as well as the communication bandwidth constraints between the virtual machine. We quantified the virtual machine cluster constructed an undirected graph. In the undirected graph, the nodes represent the virtual machine, so the weight of a node represents the value of resources, and the edges represent the communication bandwidth, so the weight of the edge represents the value of communication bandwidth. Base on the above transformations, the resources and bandwidth constrained optimization problem is transformed into the graph segmentation problem. Next we segment the undirected graph by minimum cut algorithm, and computing the matching degree of physical machines. Last we obtained the approximate solution. To validate the effectiveness of the new algorithm, we carried out extensive experiments based on the CloudSim platform.

One of the key challenges in the domain of cloud computing is task scheduling and estimation of cloud workloads for time critical applications pertaining to constrained cloud resources. While effective task scheduling is necessary for balancing the load, workload forecasting is necessary to plan in advance the requirements of cloud platforms based on previous data so as to effectively utilize cloud resources. Often it is challenging to gather sufficient information about the tasks and hence allocating the tasks to virtual machines (VMs) in the most optimal way is non-trivial. In this paper, a hybrid task scheduling approach is proposed based on evolutionary algorithms. The first approach is the amalgamation of bat and particle swarm optimization (PSO) techniques. The scheduling approach also combines the processing time preemption (PTP) approach to schedule the source intensive tasks which allows to reduce the response time of the proposed system. The second approach is a machine learning based approach employing gradient descent with momentum (GDM). The evaluation of the proposed system has been done based on the response time and mean square error of the system.

Cloud computing provides a cost-effective computing platform for big data workflows where moldable parallel computing models such as MapReduce are widely applied to meet stringent performance requirements. The granularity of task partitioning in each moldable job has a significant impact on workflow completion time and financial cost. We investigate the properties of moldable jobs and design a big-data workflow mapping model, based on which, we formulate a workflow mapping problem to minimize workflow makespan under a budget constraint in public clouds. We show this problem to be strongly NP-complete and design i) a fully polynomial-time approximation scheme (FPTAS) for a special case with a pipeline-structured workflow executed on virtual machines in a single class, and ii) a heuristic for a generalized problem with an arbitrary directed acyclic graph-structured workflow executed on virtual machines in multiple classes. The performance superiority of the proposed solution is illustrated by extensive simulation-based results in Hadoop/YARN in comparison with existing workflow mapping models and algorithms.

Advances in virtualization technology have focused mainly on strengthening the isolation barrier between virtual machines (VMs) that are co-resident within a single physical machine. At the same time, a large category of communication intensive distributed applications and software components exist, such as web services, high performance grid applications, transaction processing, and graphics rendering, that often wish to communicate across this isolation barrier with other endpoints on co-resident VMs. State of the art inter-VM communication mechanisms do not adequately address the requirements of such applications. TCP/UDP based network communication tends to perform poorly when used between co-resident VMs, but has the advantage of being transparent to user applications. Other solutions exploit inter-domain shared memory mechanisms to improve communication latency and bandwidth, but require applications or user libraries to be rewritten against customized APIs - something not practical for a large majority of distributed applications. In this paper, we present the design and implementation of a fully transparent and high performance inter-VM network loopback channel, called XenLoop, in the Xen virtual machine environment. XenLoop does not sacrifice user-level transparency and yet achieves high communication performance between co-resident guest VMs. XenLoop intercepts outgoing network packets beneath the network layer and shepherds the packets destined to co-resident VMs through a high-speed inter-VM shared memory channel that bypasses the virtualized network interface. Guest VMs using XenLoop can migrate transparently across machines without disrupting ongoing network communications, and seamlessly switch between the standard network path and the XenLoop channel. In our evaluation using a number of unmodified benchmarks, we observe that XenLoop can reduce the inter-VM round trip latency by up to a factor of 5 and increase bandwidth by a up to a factor of 6.

Usually, scientists need to execute experiments that demand high performance computing environments and parallel techniques. This is the scenario found in many bioinformatics experiments modeled as scientific workflows, such as phylogenetic and phylogenomic analyses. To execute these experiments, scientists have adopted virtual machines (VMs) instantiated in clouds. Estimating the number of VMs to instantiate is a crucial task to avoid negative impacts on the execution performance and on the financial costs with under or overestimations. Previously, the necessary number of VMs to execute bioinformatics workflows have been estimated by a GRASP heuristic and have been coupled to a Cloud-based Parallel Scientific Workflow Management System. Although this work was a step forward, this approach only provided a static dimensioning. If the characteristics of the environment change (processing capacity, network speed), this static dimensioning may not be suitable. In this way, it is of interest that the dimensioning is adjusted at runtime. To achieve this, we developed a novel framework for monitoring and dynamically dimensioning resources during the execution of parallel scientific workflows in clouds, called Dynamic Dimensioning of Cloud Computing Framework (DDC-F). We have evaluated DDC-F in real executions of bioinformatics workflows. Experiments showed that DDC-F is able to efficiently calculate the number of VMs necessary to execute bioinformatics workflows of Comparative Genomics (CG), also reducing the financial costs, when compared with other works of the related literature.

The deployment of fifth-generation (5G) and beyond 5G (B5G) networks is the ambitious objective of modern research on future mobile networks that are evolving to support computation-intensive and communication-sensitive applications. Such applications (e.g., autonomous vehicles, industrial automation, and remote surgery) impose diverse quality-of-service (QoS) requirements on the network in terms of processing, latency, reliability, and bandwidth, and will require ultra-reliable low-latency communication (URLLC), paving the way for multi-access edge computing (MEC). Our work considers a dynamic indoor B5G network in a robotic scenario where agents continuously need MEC services and migrate from one cell to another to perform their tasks in an ultra-dense cell environment. Assuming that every MEC service is a virtual machine (VM) to execute in one of the cells with the possibility of migrating the VM to another cell by paying some cost, we formalize the joint problem of (1) placing/migrating the VMs to respect their end-to-end communication latency requirements and (2) allocating their computation and communication bandwidth as a mixed-integer linear program (MILP). An MILP solver is then used to find the optimal VM placements/migrations and bandwidth allocations over a time horizon.Keywords5GMECMILPURLLCReal-time

Virtualization techniques effectively handle the growing demand for computing, storage, and communication resources in large-scale Cloud Data Centers (CDC). It helps to achieve different resource management objectives like load balancing, online system maintenance, proactive fault tolerance, power management, and resource sharing through Virtual Machine (VM) migration. VM migration is a resource-intensive procedure as VM’s continuously demand appropriate CPU cycles, cache memory, memory capacity, and communication bandwidth. Therefore, this process degrades the performance of running applications and adversely affects efficiency of the data centers, particularly when Service Level Agreements (SLA) and critical business objectives are to be met. Live VM migration is frequently used because it allows the availability of application service, while migration is performed. In this paper, we make an exhaustive survey of the literature on live VM migration and analyze the various proposed mechanisms. We first classify the types of Live VM migration (single, multiple and hybrid). Next, we categorize VM migration techniques based on duplication mechanisms (replication, de-duplication, redundancy, and compression) and awareness of context (dependency, soft page, dirty page, and page fault) and evaluate the various Live VM migration techniques. We discuss various performance metrics like application service downtime, total migration time and amount of data transferred. CPU, memory and storage data is transferred during the process of VM migration and we identify the category of data that needs to be transferred in each case. We present a brief discussion on security threats in live VM migration and categories them in three different classes (control plane, data plane, and migration module). We also explain the security requirements and existing solutions to mitigate possible attacks. Specific gaps are identified and the research challenges in improving the performance of live VM migration are highlighted. The significance of this work is that it presents the background of live VM migration techniques and an in depth review which will be helpful for cloud professionals and researchers to further explore the challenges and provide optimal solutions.