On-demand Instances Research Articles

An online bi-objective scheduling algorithm for service provisioning in cloud computing

To reduce costs and risks, service providers can purchase Infrastructure-as-a-Service (IaaS) instances from public clouds to deploy service applications and deliver job-execution services to users. However, it is challenging to make optimal instance purchasing and job scheduling decisions due to diverse instance billing methods and random service demands. In order to minimize time average service costs while ensuring quality of service (QoS), we consider that service providers purchase combinations of reserved and on-demand instances to meet the workload of their service system. Service providers need to dynamically determine the amount of on-demand and reserved instances to purchase for job executions, and then schedule jobs to run at the appropriate time. In this paper, we first propose a method to dynamically purchase reserved instances based on the historical usage data of on-demand and reserved instances. Then, we use a variable-length cycle scheduling mode for the job processing system and propose an online bi-objective scheduling algorithm based on Lyapunov optimization to schedule jobs. Even without the knowledge about the future, our online algorithm can dynamically purchase on-demand instances in an optimal way under the current number of reserved instances, and schedule jobs in varying time periods according to the arrival of jobs and optimization objectives. Based on both synthetic and real-world data, extensive experimental results show that our algorithm can achieve bi-objective optimization of minimizing time average cost while ensuring QoS.

Maximizing Profit for Cloud Brokers in The Era of Cloud Computing

Abstract: Cloud computing has experienced significant growth in recent years, with an increasing number of applications migrating to the cloud, where they can centrally manage a wide range of resources, including hardware and software. These resources are delivered over the internet as services, catering to customer demands. One of the key features of cloud computing is the pay-as-you-go model. However, due to the one-hour billing cycle and the discount structure favoring long-term users, many customers end up paying more than their actual usage. To address this issue, cloud brokers rent reserved Virtual Machines (VMs) from cloud providers at a reduced cost and offer them to users on an on-demand basis at a more affordable rate than what the cloud providers offer. Moreover, cloud brokers operate on a shorter billing cycle compared to cloud providers. This study focuses on configuring a cloud broker and determining the optimal pricing for its VMs to maximize its profits. Most cloud providers offer two primary payment options for their instances: On-Demand and Reserved Instances. With On-Demand instances, users pay for compute capacity per hour, making them suitable for short-term workloads. Reserved Instances provide significant discounts but require users to commit to long rental periods, disadvantaging short-term customers. The introduction of the cloud broker serves as a novel intermediary between cloud providers and cloud users, addressing the issues of both onehour billing cycles and the limitations on discounts for short-term users. This manuscript begins by introducing a multi-server queuing model, a revenue model, and a cost model, followed by a comprehensive analysis of the factors affecting them. It defines an optimization problem aiming to determine the optimal configuration for multi-server setups and VM pricing, all aimed at profit maximization. To address this optimization problem, a heuristic method is proposed, which combines partial derivatives and a bisection search. This method offers a practical approach to finding the optimal VM price and system scale, ultimately maximizing profit. This manuscript explores the critical role of cloud brokers in the cloud computing landscape and presents a comprehensive approach to optimizing their operations, ultimately benefitting both cloud providers and users.

Dynamic multi-objective workflow scheduling for combined resources in cloud

Cloud resource providers offer idle resources to users as spot instances. The price of the instances changes with market supply and demand, and the dynamic price can have a significant impact on workflow scheduling. In this work, we use a combination of spot and on-demand instances as the foundation cloud resource and characterize the dynamic workflow scheduling problem as a dynamic multi-objective optimization problem (DMOP), where the dynamics originate from the dynamic price of spot instances. The scheduling solution is found by considering three objectives: maximizing the reliability of the instances while minimizing the makespan and cost. In addition, we provide an enhanced MOEAD algorithm called MOEA/D-URDI that combines diversity introduction and uniform random sampling, where the uniform random sampling paradigm is used to generate the initial weight vector. The dynamic multi-objective optimization evolutionary algorithm DMOEA/D-URDI is then created by combining the method with a dynamic optimization framework. Our technique beats existing algorithms, according to experimental data based on dynamic benchmark sets and three well-known scientific procedures in terms of metrics on dynamic benchmark sets and better ensures reliability in scheduling scientific workflows while reducing makespan and cost.

Transition Waste Optimization for Coded Elastic Computing

Distributed computing, in which a resource-intensive task is divided into subtasks and distributed among different machines, plays a key role in solving large-scale problems. Coded computing is a recently emerging paradigm where redundancy for distributed computing is introduced to alleviate the impact of slow machines (stragglers) on the completion time. We investigate coded computing solutions over elastic resources, where the set of available machines may change in the middle of the computation. This is motivated by recently available services in the cloud computing industry (e.g., EC2 Spot, Azure Batch) where low-priority virtual machines are offered at a fraction of the price of the on-demand instances but can be preempted on short notice. Our contributions are three-fold. We first introduce a new concept called transition waste that quantifies the number of tasks existing machines must abandon or take over when a machine joins/leaves. We then develop an efficient method to minimize the transition waste for the cyclic task allocation scheme recently proposed in the literature (Yang et al. ISIT'19). Finally, we establish a novel solution based on finite geometry achieving zero transition wastes given that the number of active machines varies within a fixed range.

Scheduling Bag-of-Tasks in Clouds Using Spot and Burstable Virtual Machines

Leading Cloud providers offer several types of Virtual Machines (VMs) in diverse contract models, with different guarantees in terms of availability and reliability. Among them, the most popular contract models are the on-demand and the spot models. In the former, on-demand VMs are allocated for a fixed cost per time unit, and their availability is ensured during the whole execution. On the other hand, in the spot market, VMs are offered with a huge discount when compared to the on-demand VMs, but their availability fluctuates according to the cloud's current demand that can terminate or hibernate a spot VM at any time. Furthermore, in order to cope with workload variations, cloud providers have also introduced the concept of burstable VMs which are able to burst up their respective baseline CPU performance during a limited period of time with an up to 20% discount when compared to an equivalent non-burstable on-demand VMs. In the current work, we present the Burst Hibernation-Aware Dynamic Scheduler (Burst-HADS), a framework that schedules and executes tasks of Bag-of-Tasks applications with deadline constraints by exploiting spot and on-demand burstable VMs, aiming at minimizing both the monetary cost and the execution time. Based on ILS metaheuristics, Burst-HADS defines an initial scheduling map of tasks to VMs which can then be dynamically altered by migrating tasks of a hibernated spot VM or by performing work-stealing when VMs become idle. Performance results on Amazon EC2 cloud with different applications show that, when compared to a solution that uses only regular on-demand instances, Burst-HADS reduces the monetary cost of the execution and meet the application deadline even in scenarios with high spot hibernation rates. It also reduces the total execution time when compared to a solution that uses only spot and non-burstable on-demand instances.

Learning to make auto-scaling decisions with heterogeneous spot and on-demand instances via reinforcement learning

Designing auto-scaling frameworks using spot and on-demand instances while considering their heterogeneity, can help Software-as-a-Service (SaaS) providers provide services with high availability to meet dynamic workloads and achieve significant cost savings. However, designing such an auto-scaling framework is difficult due to the lack of prior knowledge of the cloud. In this work, we propose an algorithm called SpotRL to solve the auto-scaling problem using heterogeneous spot and on-demand instances. Reinforcement learning (RL) approaches have been shown to be able to make effective decisions in highly dynamic environments, as they can learn step-by-step and find solutions without prior knowledge. SpotRL uses an RL-based approach for the scaling of heterogeneous spot instances. In the complex cloud environment, the training speed of RL agents is generally slow. Considering this issue, we use a multi-agent approach to decompose tasks to help agents learn faster. To reduce the negative impact of low service availability due to agents’ random explorations as they interact with the cloud environment, SpotRL uses a passive approach for the scaling of heterogeneous on-demand instances. Our experimental results show that the SpotRL approach can significantly reduce the deployment cost of SaaS providers while complying with high service availability.

DOLL

Most large-scale ML implementations scale to large amounts of data by utilizing multiple servers or virtual machines (VMs) that iteratively compute model updates on local data that are periodically synchronized. Due to the complexity of managing the resulting computing infrastructure, many companies run their ML jobs on external cloud providers' servers. However, cloud resources can be expensive, particularly for large ML jobs with long runtimes. A particularly popular method to limit the costs of training ML jobs is to utilize preemptible cloud instances. These may be interrupted at the cloud provider's discretion, but they are significantly (up to 90%) cheaper than conventional on-demand instances. Most studies of these ML methods, however, assume the availability of large datasets at training time. In practice, training data may arrive at irregular intervals and models may be trained online as new data samples arrive, e.g., when monitoring data from IoT sensors. While some software frameworks like Apache Kafka can feed online data arrivals to ML algorithms, they provide little insight into the resulting costs of ML training. We extend prior work on provisioning preemptible instances to analyze available pools of data in order to run online ML on incoming datastreams, which presents new challenges due to the need to carefully handle data arrivals. We design, analyze, and optimize DOLL, which to the best of our knowledge is the first system that provides provable performance guarantees for Distributed OnLine Learning over preemptible instances. Research Challenges and Our Contributions: When pools of data are readily available, the bottleneck to distributed ML training often lies in the time required for each VM to compute its model updates. In our scenario, however, the arrival rate of incoming data may also bottleneck data processing. An intuitive strategy would then be for each VM to process each data point as it arrives. However, since arrivals at different VMs may not be coordinated, synchronizing the model parameters at each VM between data arrivals may introduce additional delays, while asynchronous SGD methods can lead to slow convergence [1]. DOLL uses a batching and grouping process to limit the synchronization delay, which naturally realizes traditional mini-batch SGD so as to provide provable model convergence guarantees. Handling online data arrivals becomes particularly challenging when we use preemptible instances to compute model updates. Existing methods utilizing preemptible instances for ML jobs largely focus on mitigating training interruptions [2] and their effects on model convergence [3]. When used on datastreams, we face an additional challenge of interruptions pausing the data arrival process, which impedes the rate at which we can compute model updates and thus model convergence. Thus, one should ensure that preemptions do not happen "too often," e.g., by computing some updates on on-demand instances. Our work is the first to optimize the number of preemptible VMs used and demonstrate that we can meet ML convergence guarantees.

A Cost-Effective Framework for Running Industrial Big Data Analysis Applications in Public Clouds

Nowadays, the improvement of the data acquisition capability of the Industrial Internet of Things (IIoT) systems has brought about higher data throughput. Users can deploy industrial big data analysis applications on cloud platforms in a pay-as-you-go way to deal with the unstable data generation and data analysis workloads in the IIoT scenario. To procure stable and flexible computing resources, as a traditional pay-as-you-go cloud instance procurement option, on-demand instances are widely used, but their expensive prices also significantly increase users’ cost burden. To reduce the data analysis cost, in this article, we establish a per-job cost-effective framework adopting spot instances, on-demand instances, and cloud storage for industrial big data analysis applications. In our framework, we take advantage of spot instances, which are computing instances provided at low prices under a pay-as-you-go model, to achieve cost savings. However, using spot instances carries the risk of being interrupted. Therefore, we propose to use a checkpointing mechanism to back up intermediate results to cloud storage to reduce the potential loss caused by spot instance interruptions. Considering the time sensitivity of industrial big data analysis applications, we use on-demand instances as alternative computing resources after spot instance interruptions to ensure that users’ jobs can be completed without high time latencies. Evaluation results show that our framework can achieve cost savings as well as minimize time latencies for users’ jobs.

An online algorithm for optimally releasing multiple on-demand instances in IaaS clouds

With the popularity of cloud computing increasing rapidly in recent years, the use of cloud, represented by IaaS and SaaS, is being embraced by more and more users. One of the flexibility of the cloud lies in its pay-as-you-go usage model, which allows users to purchase and release cloud instances on their own demand, reducing the possible financial loss caused by wasted resources. For a SaaS provider that uses the pay-as-you-go payment model to purchase cloud instances, it is important to make a reasonable decision on when to release as many as idle on-demand cloud instances to achieve cost savings when the number of incoming user demands is in a declining phase, taking into account the cost of the start-up time to acquire new cloud instances and the penalty cost that may be incurred while SaaS users wait. In order to make optimal decisions when there is not enough knowledge to predict the future trend of incoming demands, we propose an online instance releasing algorithm which can effectively help SaaS providers to reduce the cost when using on-demand instances. Through theoretical analysis we show our online algorithm can achieve a competitive ratio of less than 2 for a variety of penalty functions. Our extensive simulation experiments based on both real Google workload data and simulated demand sequences demonstrate that the proposed online algorithm is stable and efficient.

A Random Online Algorithm for Reselling Reserved IaaS Instances in Amazon's Cloud Marketplace

When running applications in IaaS (Infrastructure as a Service) cloud platforms like Amazon EC2, cloud users can choose to purchase reserved instances instead of on-demand ones to save cost. However, if there are few workload arrivals after instance reservations, it will cause a waste of cost due to the idle reserved instances. At present, Amazon EC2 has launched a reserved instance market, in which cloud users can resell their idle reserved instances to avoid wastes. However, for IaaS users, it is difficult to make the optimal decisions on whether to resell idle reserved instances when they do not know the arrival situation of workloads in the future. This is because after the idle reserved instances are sold, if there arrive new workloads, additional cost for launching new reserved or on-demand instances will be incurred. In this paper, we propose a random online reselling algorithm to help cloud users make reasonable decisions on whether to sell their idle instances, while without requiring the knowledge of the future workload arrivals. We theoretically prove that our algorithm can achieve a bounded competitive ratio. Besides, we conduct extensive experiments using the real-world workload obtained from public clouds to validate our algorithm.

A Lyapunov optimization-based online scheduling algorithm for service provisioning in cloud computing

For economically provisioning services, service providers can purchase combinations of reserved and on-demand instances from Infrastructure-as-a-Service (IaaS) clouds. These different billing options bring challenges to service providers on how to purchase instances optimally to ensure quality of service (QoS) while maximizing revenue. In order to ensure QoS and maximize average revenue, a service provider can purchase fixed number of reserved instances to meet the base workload of its service system, and then dynamically purchase on-demand instances to satisfy the service requests beyond its base workload. However, the random, dynamic, and uncertain nature of service request arrivals makes it difficult for service providers to determine the optimal number of on-demand instances to be purchased dynamically. To solve the above problems, in this paper, we propose an online scheduling algorithm based on the theory of Lyapunov optimization and explore the concept of capacity region and queue stability to ensure QoS. Our online algorithm dynamically purchases on-demand instances and schedules jobs to run according to the bidding price of users without requiring the future workload information. Through our proposed online scheduling algorithm, service providers can both ensure QoS and maximize average revenue, and make a trade-off between them. Simulation results show that our algorithm can effectively achieve bi-objective optimization of maximizing time average revenue as well as ensuring QoS.

Laniakea@ReCaS: exploring the potential of customisable Galaxy on-demand instances as a cloud-based service

BackgroundImproving the availability and usability of data and analytical tools is a critical precondition for further advancing modern biological and biomedical research. For instance, one of the many ramifications of the COVID-19 global pandemic has been to make even more evident the importance of having bioinformatics tools and data readily actionable by researchers through convenient access points and supported by adequate IT infrastructures. One of the most successful efforts in improving the availability and usability of bioinformatics tools and data is represented by the Galaxy workflow manager and its thriving community. In 2020 we introduced Laniakea, a software platform conceived to streamline the configuration and deployment of “on-demand” Galaxy instances over the cloud. By facilitating the set-up and configuration of Galaxy web servers, Laniakea provides researchers with a powerful and highly customisable platform for executing complex bioinformatics analyses. The system can be accessed through a dedicated and user-friendly web interface that allows the Galaxy web server’s initial configuration and deployment.Results“Laniakea@ReCaS”, the first instance of a Laniakea-based service, is managed by ELIXIR-IT and was officially launched in February 2020, after about one year of development and testing that involved several users. Researchers can request access to Laniakea@ReCaS through an open-ended call for use-cases. Ten project proposals have been accepted since then, totalling 18 Galaxy on-demand virtual servers that employ ~ 100 CPUs, ~ 250 GB of RAM and ~ 5 TB of storage and serve several different communities and purposes. Herein, we present eight use cases demonstrating the versatility of the platform.ConclusionsDuring this first year of activity, the Laniakea-based service emerged as a flexible platform that facilitated the rapid development of bioinformatics tools, the efficient delivery of training activities, and the provision of public bioinformatics services in different settings, including food safety and clinical research. Laniakea@ReCaS provides a proof of concept of how enabling access to appropriate, reliable IT resources and ready-to-use bioinformatics tools can considerably streamline researchers’ work.

Optimized Portfolio Contracts for Bidding the Cloud

Amazon EC2 provides two most popular pricing schemes–i) the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">costly</i> on-demand instance where the job is guaranteed to be completed, and ii) the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">cheap</i> spot instance where a job may be interrupted. We consider a user can select a combination of on-demand and spot instances to finish a task. Thus he needs to find the optimal bidding price for the spot-instance, and the portion of the job to be run on the on-demand instance. We formulate the problem as an optimization problem and seek to find the optimal solution. We consider three bidding strategies: one-time requests with expected guarantee, one-time requests with penalty for incomplete job and violating the deadline, and persistent requests. Even without a penalty on incomplete jobs, the optimization problem turns out to be non-convex. Nevertheless, we show that the portion of the job to be run on the on-demand instance is at most half. If the job has a higher execution time or smaller deadline, the bidding price is higher and vice versa. Additionally, the user never selects the on-demand instance if the execution time is smaller than the deadline. The numerical results illustrate the sensitivity of the effective portfolio to several of the parameters involved in the model. Our empirical analysis on the Amazon EC2 data shows that our strategies can be employed on the real instances, where the expected total cost of the proposed scheme decreases over 45 percent compared to the baseline strategy.

Hybrid Resource Provisioning for Cloud Workflows with Malleable and Rigid Tasks

In cloud computing, reserved and on-demand instances are generally provided by service providers. Hybridization of the two alternatives can considerably save costs when renting resources from the cloud. However, it is a big challenge to determine the appropriate amount of reserved and on-demand resources in terms of users’ requirements. In this paper, the workflow scheduling problem with both reserved and on-demand instances is considered. The objective is to minimize the total rental cost under deadline constrains. The considered problem is mathematically modeled. A multiple sequence-based earliest finish time method is proposed to construct schedules for the workflows. Four different rules are used to generate initial task allocation sequences. Types and quantities of resources are determined by a free time block-based schedule construction mechanism. New sequences are generated by a variable neighborhood search method. Experimental and statistical analyses and results demonstrate that the proposed algorithm algorithm generates considerable cost savings when compared to the algorithms with only on-demand or reserved instances.

A Two-Phase Cloud Resource Provisioning Algorithm for Cost Optimization

Cloud computing is a new computing paradigm to deliver computing resources as services over the Internet. Under such a paradigm, cloud users can rent computing resources from cloud providers to provide their services. The goal of cloud users is to minimize the resource rental cost while meeting the service requirements. In reality, cloud providers often offer multiple pricing models for virtual machine (VM) instances, including on-demand and reserved pricing models. Moreover, the workload of cloud users varies with time and is not known a priori. Therefore, it is challenging for cloud users to determine the optimal cloud resource provisioning. In this paper, we propose a two-phase cloud resource provisioning algorithm. In the first phase, we formulate the resource reservation problem as a two-stage stochastic programming problem, and solve it by the sample average approximation method and the dual decomposition method. In the second phase, we propose a hybrid ARIMA-Kalman model to predict the workload, and determine the number of on-demand instances based on the predicted workload. The effectiveness of the proposed two-phase algorithm is evaluated using a real-world workload trace and Amazon EC2’s pricing models. The simulation results show that the proposed algorithm can significantly reduce the operational cost while guaranteeing the service level agreement (SLA).

Cloud spot instance price prediction using kNN regression

Cloud computing can provide users with basic hardware resources, and there are three instance types: reserved instances, on-demand instances and spot instances. The price of spot instance is lower than others on average, but it fluctuates according to market demand and supply. When a user requests a spot instance, he/she needs to give a bid. Only if the bid is not lower than the spot price, user can obtain the right to use this instance. Thus, it is very important and challenging to predict the price of spot instance. To this end, we take the most popular and representative Amazon EC2 as a testbed, and use the price history of its spot instance to predict future price by building a k-Nearest Neighbors (kNN) regression model, which is based on our mathematical description of spot instance price prediction problem. We compare our model with Linear Regression (LR), Support Vector Machine Regression (SVR), Random Forest (RF), Multi-layer Perception Regression (MLPR), gcForest, and the experiments show that our model outperforms the others.

Constructing Reliable Computing Environments on Top of Amazon EC2 Spot Instances

Cloud provider Amazon Elastic Compute Cloud (EC2) gives access to resources in the form of virtual servers, also known as instances. EC2 spot instances (SIs) offer spare computational capacity at steep discounts compared to reliable and fixed price on-demand instances. The drawback, however, is that the delay in acquiring spots can be incredible high. Moreover, SIs may not always be available as they can be reclaimed by EC2 at any given time, with a two-minute interruption notice. In this paper, we propose a multi-workflow scheduling algorithm, allied with a container migration-based mechanism, to dynamically construct and readjust virtual clusters on top of non-reserved EC2 pricing model instances. Our solution leverages recent findings on performance and behavior characteristics of EC2 spots. We conducted simulations by submitting real-life workflow applications, constrained by user-defined deadline and budget quality of service (QoS) parameters. The results indicate that our solution improves the rate of completed tasks by almost 20%, and the rate of completed workflows by at least 30%, compared with other state-of-the-art algorithms, for a worse-case scenario.

Multi-stage stochastic programming models for provisioning cloud computing resources

We focus on the resource provisioning problem of a cloud consumer from an Infrastructure-as-a-Service type of cloud. The cloud provider offers two deployment options, which can be mixed and matched as appropriate. Cloud instances may be reserved for a fixed time period in advance at a smaller usage cost per hour but require a full commitment and payment for the entire contract duration. In contrast, on-demand instances reflect a pay-as-you-go policy at a premium. The trade-off between these two options is rooted in the inherent uncertainty in demand and price and makes it attractive to complement a base reserved capacity with on-demand capacity to hedge against the spikes in demand. This paper provides several novel multi-stage stochastic programming formulations to enable a cloud consumer to handle the cloud resource provisioning problem at a tactical level. We first formulate the cloud resource provisioning problem as a risk-neutral multi-stage stochastic program, which serves as the base model for further modeling variants. In our second set of models, we also incorporate a certain concept of system reliability. In particular, chance constraints integrated into the base formulation require a minimum service level met from reserved capacity, provide more visibility into the future available capacity, and smooth out expensive on-demand usage by hedging against possible demand fluctuations. An extensive computational study demonstrates the value of the proposed models by discussing computational performance, gleaning practical managerial insights from the analysis of the solutions of the proposed models, and quantifying the value of the stochastic solutions.

Evolutionary Multi-Objective Workflow Scheduling for Volatile Resources in the Cloud

The cloud has been widely used as a distributed computing platform for running scientific workflow applications. Most of the cloud providers encourage the use of their underutilized resources as spot instances for much cheaper prices compared with common resources as on-demand instances, however, the promise of lower costs for resources results in the volatility such that spot instances can be interrupted at any time by cloud providers. Many workflow scheduling algorithms have been proposed to deal with volatile resources. In this article, we consider the two most important features of the volatile resources namely fulfillment and interruption rates to fully model the instability of the cloud infrastructure. Subsequently, we propose a novel evolutionary multi-objective workflow scheduling approach to generate a set of trade-off solutions that outperform state-of-the-art algorithms in both makespan and economic costs. In addition, we explore the fluctuation of makespan and costs for our obtained schedules under different levels of fulfillment and interruption rates. Experimental results with the five well-known real-world workflows demonstrate that our evolutionary multi-objective workflow scheduling algorithm is competitive in terms of makespan and cost compared with state-of-the-art on-demand scheduling techniques.

Virtual Machine Trading in Public Clouds

By obtaining virtual machines (VMs) from infrastructure providers (InPs) according to the demand in public cloud services, service providers (SPs) can elastically provide network services to users. As the charging methods of VMs, reserved instance (RI) and on-demand instance (ODI) are widely used. For InPs, RI is more desirable than ODI thanks to easiness of estimating long-term revenue, risk aversion of occurring idle VM resources, and reduction of charging cost. In this paper, to improve the ratio of RI in VMs prepared by an InP, we propose VM trading methods in which idle RI of SPs with VM demand falling below the number of contracted RI are transferred to SPs with VM demand exceeding the number of contracted RI. As the VM trading mechanisms, we investigate two approaches: RI with self-help effort (RISE) and RI with mutual aid (RIMA). Through numerical evaluation using the demand pattern of commercial VoD service, we show that the proposed VM trading methods decrease the number of VMs required for ODI by about 50% to 100% and increase the ratio of RI by about 15% to 85%.