Prototype Cloud Research Articles

Distributed Cloud Computing: Architecture, Enabling Technologies, and Open Challenges

Conventional cloud computing, where compute, storage, and networking resources reside in one or a few centralized data centers, has become unable to meet the stringent latency requirements of new applications. Along with that, the rapid development of 5 G network introduces the trend of network cloudification and network service provisioning according to cloud service models. Therefore, the emerging distributed cloud model represents an evolution from conventional centralized cloud into geographically dispersed cloud computing services located as per application needs. In this article, we attempt to sketch a big picture of distributed cloud. We first interpret the concept of distributed cloud computing. Then, we describe an architecture of the distributed cloud along with the enabling technologies. We also carry out a case study on service deployment and discovery in a 3-layer distributed cloud prototype and provide a preliminary assessment of service discovery time. Finally, we discuss open research challenges towards distributed cloud.

An Improved Explainable Point Cloud Classifier (XPCC)

Classification of objects from 3-D point clouds has become an increasingly relevant task across many computer-vision applications. However, few studies have investigated explainable methods. In this article, a new prototype-based and explainable classification method called eXplainable point cloud classifier (XPCC) is proposed. The XPCC method offers several advantages over previous explainable and nonexplainable methods. First, the XPCC method uses local densities and global multivariate generative distributions. Therefore, the XPCC provides comprehensive and interpretable object-based classification. Furthermore, the proposed method is built on recursive calculations, thus, is computationally very efficient. Second, the model learns continuously without the need for complete retraining and is domain transferable. Third, the proposed XPCC expands on the underlying learning method explainable deep neural networks (xDNN), and is specific to 3-D. As such, the following three new layers are added to the original xDNN architecture: 1) the 3-D point cloud feature extraction, 2) the global compound prototype weighting, and 3) the SoftMax function. Experiments were performed with the ModelNet40 benchmark, which demonstrated that XPCC is the only one to increase classification accuracy relative to the base algorithm when applied to the same problem. In addition, this article proposes a novel prototype-based visual representation that provides model- and object-based explanations. The prototype objects are superimposed to create a prototypical class representation of their data density within the feature space, called the compound prototype cloud. They allow a user to visualize the explainable aspects of the model and identify object regions that contribute to the classification in a human-understandable way.

BRAS : Development of Prototype Cloud Model for Failure Recovery Management by Using Backup Resource Allocation Strategy

The main objective of any technology is to give good service, availability and reliability to the end user along with protection and cost feasibility. Cloud computing is one of such technology which can be pay-as-you-go model. Computational resources and backup resources are the one of the issues. When multiple Physical Machines (PM) interacting to cloud to have respective services there may be a chances of failures that spoils the guaranteed services by the providers. In this paper we tried to elaborate these issues by developing a prototype cloud model for failure recovery management with the information of backup resource allocation strategy. We proposed an advanced open stack method based on BRAS. We also conducted a survey on BRAS to give better model. In this paper we focussed more on availability analytical model how its work for BRAS. We also covered some of case studies with yielding results for better understanding of the model. However one of the essential pitfalls in cloud computing is related to optimizing the property being allocated. Because of the distinctiveness of the model, useful resource allocation is achieved with the aim of minimizing the prices associated with it. The specific traumatic conditions of useful resource allocation are meeting customer desires and application requirements.

Technique for Balanced Load Balancing in Cloud Computing Environment

Resource sharing by means of load balancing in cloud computing environments helps for efficient utilization of cloud resources and higher overall throughput. However, implementation of poor load balancing algorithms may cause some virtual machines starving for additional cloud resources. Employing meagre crafted mechanism for priority-oriented load balancing may leave low-level priority virtual machines starving. We suggest an improved resource sharing mechanism for load balancing in the cloud computing environments. The suggested mechanism helps to provide efficient load balancing by avoiding starvation. In order to cater efficient load balancing, the proposed resource sharing technique takes respective virtual machines’ priority levels into consideration. An implementation of the suggested load balancing algorithm in cloud environment provides reduction in waiting time of the starving virtual machines which are looking for additional resources in cloud platform. The implementation of our proposed algorithm has been deployed on a prototype cloud computing infrastructure testbed established with open source software OpenStack. The prototype cloud testbed is supported in backend by the open source CentOS Linux operating system’s minimal setup. Experimental results of proposed load balancing mechanism in the prototype cloud computing infrastructure setup designate reduction in the waiting time of overloaded starving virtual machines. The proposed mechanism is beneficial to accomplish priority-oriented and starvation free resource sharing for load balancing in cloud computing environments. In future, the proposed technique can be further enhanced for implementing load balancing in collaborated cloud computing environments.

E-Render: Enabling UHD-Quality Cloud Gaming Through Edge Rendering

Cloud gaming services enable users to enjoy various high-quality computer games without installing them in their devices which may have poor resources. However, it consumes high bandwidth in backbone networks as game videos have to be constantly streamed to client devices, which can be a hindering factor for providing Ultra High Definition (UHD)-quality cloud gaming services whose demands are ever increasing. In this paper, we present a novel cloud gaming architecture, <i>E-Render</i>, in which the graphics engine is separated from the cloud game server and is deployed in the edge for rendering. In the architecture, since video frames are streamed from the edge server to the client device via highbandwidth subscriber networks, we can greatly save the backbone network bandwidth. To realize it, however, we need to effectively reduce the amount and latency of graphics commands transfer from the game engine in the cloud to the graphics engine in the edge server. E-render solves the problem by using novel techniques including multi-granular command deduplication and asynchronous buffer binding. We implement a prototype cloud gaming platform based on the E-Render architecture, and evaluate its performance using four real world games. The evaluation results confirm that E-Render can provide UHD-quality cloud gaming services.

A Path Relinking Method for the Joint Online Scheduling and Capacity Allocation of DL Training Workloads in GPU as a Service Systems

The Deep Learning (DL) paradigm gained remarkable popularity in recent years. DL models are used to tackle increasingly complex problems, making the training process require considerable computational power. The parallel computing capabilities offered by modern GPUs partially fulfill this need, but the high costs related to GPU as a Service solutions in the cloud call for efficient capacity planning and job scheduling algorithms to reduce operational costs via resource sharing. In this work, we jointly address the online capacity planning and job scheduling problems from the perspective of cloud end-users. We present a Mixed Integer Linear Programming (MILP) formulation, and a path relinking-based method aiming at optimizing operational costs by (i) rightsizing Virtual Machine (VM) capacity at each node, (ii) partitioning the set of GPUs among multiple concurrent jobs on the same VM, and (iii) determining a due-date-aware job schedule. An extensive experimental campaign attests the effectiveness of the proposed approach in practical scenarios: costs savings up to 97&#x0025; are attained compared with first-principle methods based on, e.g., Earliest Deadline First, cost reductions up to 20&#x0025; are obtained with respect to a previously proposed Hierarchical Method and up to 95&#x0025; against a dynamic programming-based method from the literature. Scalability analyses show that systems with up to 100 nodes and 450 concurrent jobs can be managed in less than 7 seconds. The validation in a prototype cloud environment shows a deviation below 5&#x0025; between real and predicted costs.

Runtime Verification of Business Cloud Workflow Temporal Conformance

Business cloud workflows are often designed with multiple time constraints for timely response to business requests. To ensure on-time completion of workflow instances, workflow temporal conformance state needs to be constantly monitored and verified at runtime. Considering the fact that there are a large number of workflow instances running in a parallel fashion in many business scenarios, conventional verification approaches for time-related properties using such as temporal logic or timed Petri nets are not feasible due to the limitation of low efficiency at runtime. To address this issue, we propose a new approach to automated runtime verification of temporal conformance for parallel workflow instances in a cloud environment. Instead of using response time to verify temporal conformance of every single workflow as in conventional strategies, workflow throughput is employed as the performance measurement to efficiently monitor a large number of parallel workflow instances. On this basis we present a novel conformance verification strategy. This strategy considers the effect of time delay propagation in the cloud workflow systems to accurately verify workflow runtime temporal conformance. Our verification strategy is implemented in a prototype cloud workflow system and the evaluation results show that it outperforms the state-of-the-art workflow temporal verification strategy.

Experiment and Availability Analytical Model of Cloud Computing System Based on Backup Resource Sharing and Probabilistic Protection Guarantee

A probabilistic protection guarantee enables a cloud provider to improve the availability of their cloud computing system in a cost-efficient manner. A backup resource allocation strategy based on the probabilistic protection guarantee reduces the total amount of required backup computation resources by allowing multiple virtual machines to share the same backup computation resources of a physical machine. There have been no experimental studies that investigate the impact of applying the probabilistic protection guarantee to a cloud computing framework in real use. This paper presents an experiment of failure recovery on the cloud computing system in which the backup computation resources are shared by multiple virtual machines. We implement a prototype cloud system by using the OpenStack framework to demonstrate the failure recovery scenario according to the backup resource allocation strategy. We develop an availability analytical model for the backup resource allocation strategy. Based on the analytical model, we present case studies which derive the availability of cloud system by using the measurement results of the experiment.

A Deep Learning Trained Clear-Sky Mask Algorithm for VIIRS Radiometric Bias Assessment

Clear-sky mask (CSM) is a crucial influence on the calculating accuracy of the sensor radiometric biases for spectral bands of visible, infrared, and microwave regions. In this study, a fully connected deep neural network (FCDN) was proposed to generate CSM for the Visible Infrared Imaging Radiometer Suite (VIIRS) onboard Suomi National Polar-Orbiting Partnership (S-NPP) and NOAA-20 satellites. The model, well-trained by S-NPP data, was used to generate both S-NPP and NOAA-20 CSMs for the independent data, and the results were validated against the biases between the sensor observations and Community Radiative Transfer Model (CRTM) calculations (O-M). The preliminary result shows that the FCDN-CSM model works well for identifying clear-sky pixels. Both O-M mean biases and standard deviations were comparable with the Advance Clear-Sky Processor over Ocean (ACSPO) and were significantly better than a prototype cloud mask (PCM) and the case without a clear-sky check. In addition, by replacing CRTM brightness temperatures (BTs) with the atmosphere air temperature and water vapor contents as input features, the FCDN-CSM exhibits its potential to generate fast and accurate VIIRS CSM onboard follow-up Joint Polar Satellite System (JPSS) satellites for sensor calibration and validation before the physics-based CSM is available.

Workload Consolidation for Cloud Data Centers with Guaranteed QoS Using Request Reneging

Cloud data centers are widely employed to offer reliable cloud services. However, low resource utilization and high power consumption have been great challenges for cloud providers. Moreover, the rapid increase in demand for affordable cloud services magnifies the obstacles for proficient resource management policies. In this paper, we investigate how to improve resource utilization and power consumption in cloud data centers when delivering services with statistically guaranteed Quality of Service (QoS). We assume that the service provider hosts different types of services, each of which has request classes with different QoS requirements. Different from the traditional approaches that distribute workloads with different QoS levels on different Virtual Machines (VMs), we introduce an approach to pack requests of the same service type, even with different QoS requirements, into the same VM, and to remove potential failure requests in time to improve resource usage and energy cost. We formally prove that our algorithm can statistically guarantee QoS conditions in terms of deadline miss ratios. We develop a cloud prototype to empirically validate our proposed methods and algorithm. Our experimental results demonstrate that our approach can significantly outperform other traditional approaches in terms of QoS guarantees, power consumption, resource demand and electricity cost.

A practical cross-datacenter fault-tolerance algorithm in the cloud storage system

The fault-tolerance property in most cloud storage systems are designed within the scale of a single datacenter. The single datacenter as a whole may be unreachable or crashed due to severe problems, such as broken network links, power supply interruptions, and natural disasters, etc. Therefore, the design of an effective cross-datacenter fault-tolerant storage system is important to protect data security in the cloud. However, building a cross-datacenter fault-tolerant system faces great challenges, such as high latency, low throughput, high costs of bandwidth resources between datacenters. In this paper, we propose a practical cross-datacenter fault-tolerant (CDFT) algorithm in the cloud storage system. Our fault-tolerant algorithm design considers the difficult tradeoffs among fault tolerance, latency, throughput, network and storage costs. We propose the Domain Fault Codes (DFC) and the topology-aware scheduling techniques, which can tolerate the whole datacenter breakdown. We implemented the DFC-CDFT algorithm in a prototype cloud storage system. The experimental results showed that the proposed DFC-CDFT algorithm can effectively recover data blocks from the single datacenter failure while achieves low storage and bandwidth costs.

LEARNING SOLUTIONS WITH CLOUD TECHNOLOGIES

This paper presents a new approach to the organization of ICT training courses on the basis of the designed and developed a private training cloud prototype. The developed prototype was built on the Microsoft System Center 2012 resources, which allowed consolidating high-performance computing tools, combined different classes of storage devices and provided these resources on demand. We describe the implementation of SaaS private cloud model for ICT user courses and PaaS model - for ICT programming courses.

Software Defined Networking challenges and future direction: A case study of implementing SDN features on OpenStack private cloud

Cloud computing provides services on demand instantly, such as access to network infrastructure consisting of computing hardware, operating systems, network storage, database and applications. Network usage and demands are growing at a very fast rate and to meet the current requirements, there is a need for automatic infrastructure scaling. Traditional networks are difficult to automate because of the distributed nature of their decision making process for switching or routing which are collocated on the same device. Managing complex environments using traditional networks is time-consuming and expensive, especially in the case of generating virtual machines, migration and network configuration. To mitigate the challenges, network operations require efficient, flexible, agile and scalable software defined networks (SDN). This paper discuss various issues in SDN and suggests how to mitigate the network management related issues. A private cloud prototype test bed was setup to implement the SDN on the OpenStack platform to test and evaluate the various network performances provided by the various configurations.

Lightweight Management of Resource-Constrained Sensor Devices in Internet of Things

It is predicted that billions of intelligent devices and networks, such as wireless sensor networks (WSNs), will not be isolated but connected and integrated with computer networks in future Internet of Things (IoT). In order to well maintain those sensor devices, it is often necessary to evolve devices to function correctly by allowing device management (DM) entities to remotely monitor and control devices without consuming significant resources. In this paper, we propose a lightweight RESTful Web service (WS) approach to enable device management of wireless sensor devices. Specifically, motivated by the recent development of IPv6-based open standards for accessing wireless resource-constrained networks, we consider to implement IPv6 over low-power wireless personal area network (6LoWPAN)/routing protocol for low power and lossy network (RPL)/constrained application protocol (CoAP) protocols on sensor devices and propose a CoAP-based DM solution to allow easy access and management of IPv6 sensor devices. By developing a prototype cloud system, we successfully demonstrate the proposed solution in efficient and effective management of wireless sensor devices.

Simulation as a cloud service for short-run high throughput industrial print production using a service broker architecture

Evaluating end-to-end systems is uniquely challenging in industrial/commercial printing due to a large number of equipment combinations and customization needed for each customer and application. Moreover any mismatch in capacities may render multi-million dollar investments to zero returns on investment. Simulation can help foresee changes on the shop floor when demand changes. Providing a library of components that can be assembled together is the usual approach used by many simulation vendors which still leaves a simulation engineer in the loop to make it usable. We detail our experiences on implementing a prototype (private) cloud service using service broker architecture and a dynamic model generator. The service broker handles the heterogeneity associated with demand and equipment configurations whereas the dynamic model generator customizes a generic model based on inputs from the user. This helps avoiding rewiring of simulation models on each engagement. The schema and the necessary front and back-end codes all reside in the cloud and, therefore, users pay on a per use basis without worrying about the upgrade/update of software at their end. The service supports multi-tenancy which results in low costs per user and provides sharing of resource information yet restricting access to proprietary workflows and policies. A typical run costs a very small amount, which is affordable for even small-sized PSPs. We show the utility of our work in the context of educational book publishing to evaluate equipment changes needed when the current lumpy order demand stream changes to a highly fragmented demand stream. We also discuss how our work can be extended to several other domains such healthcare, transportation, 3D printing.

Clouding Technologies for Training

This paper presents a new approach to the organization of computer training on the basis of the private training cloud prototype designed and developed by the staff of Software Development Department at ITMO University. Modern IT technologies were used to create a private training cloud prototype which made it possible to consolidate high-performance computing tools, combine different classes of storage devices and offer these resources to both educators and trainees on demand.

An Experimental Validation of Public Cloud Mobile Banking

Currently, financial institutions incur huge expenditure to implement and maintain mobile banking (m- Banking) solutions and this cost is bound to rise significantly, as more customers subscribe to m-Banking services. Cloud computing has potential to facilitate reduced cost, high scalability and a variable cost structure that could guarantee cheaper, reliable and sustainable m-Banking in the long term. While the adoption of organizational private clouds seems natural for banks because of the sensitive nature of banking transactions, some have argued for the adoption of public clouds as a better alternative, despite concerns on issues such as trust, security and privacy. However, there is lack of sufficient empirical evidence in the literature on the suitability of public clouds for m- Banking. Hence, this study presents an investigation of the use of public cloud for m-Banking. A prototype cloud- based m-Banking application was developed using a public platform-as-a-service (Paas) cloud model, which was evaluated for usability and robustness in a controlled experiment. The evaluation result shows that m-Banking on public cloud is viable, if the cloud-based application is sufficiently robust and usable. The result also indicates that m-Banking services on public cloud are suitable for adoption by the banking industry.

Adaptive Algorithm for Minimizing Cloud Task Length with Prediction Errors

Compared to traditional distributed computing like grid system, it is non-trivial to optimize cloud task’s execution performance due to its more constraints like user payment budget and divisible resource demand. In this paper, we analyze in-depth our proposed optimal algorithm minimizing task execution length with divisible resources and payment budget: 1) We derive the upper bound of cloud task length, by taking into account both workload prediction errors and hostload prediction errors. With such state-of-the-art bounds, the worst-case task execution performance is predictable, which can improve the quality of service in turn. 2) We design a dynamic version for the algorithm to adapt to the load dynamics over task execution progress, further improving the resource utilization. 3) We rigorously build a cloud prototype over a real cluster environment with 56 virtual machines, and evaluate our algorithm with different levels of resource contention. Cloud users in our cloud system are able to compose various tasks based on off-the-shelf web services. Experiments show that task execution lengths under our algorithm are always close to their theoretical optimal values, even in a competitive situation with limited available resources. We also observe a high level of fair treatment on the resource allocation among all tasks.

R-Wingu, the ‘Big Data’ Analytic Framework: a Solution to Intelligent Correlation of Research Output in a Private Cloud Prototype for Seamless Research Ontologies

‘Internet of Things’ is choking the world with Zettabyte scale data rate of which traditional computing can neither store nor process. Outputs of various kinds of Research endeavors are held in different media over the years and cannot be made to “talk to each other” to exploit each other for beneficial relative information. The prime objective is to find a platform that can easily accommodate Cloud technology which can elastically handle the big data concept. In this study a private cloud is built using Ubuntu and Eucalyptus open source software on two quad processor machines with 8GB Ram. Using Apache Flume and Hadoop analytics, Big Data obtained from Research outputs is mounted on the R-Wingu framework with capacity to data mine unstructured, unrelated data and relate the components intelligently and may be leased as Database as a Service(DaaS). Once disparate data streams are accessible in real time, in one place and a consistent fashion, data suddenly becomes much more powerful and decisions become that much more impactful. Surveys are conducted across research communities with an aim of isolating research Gaps through various statistical tools. Sample results are discussed alongside previous studies’ outputs. Research Gaps filled are compared by poll results conducted with exiting outcomes. Establishments are urged to embrace Big Data approach for competitive advantage.

Cloud covering denoising through image fusion

This paper presents a solution to the cloud removal problem, based in a recently developed image fusion methodology consisting in applying a 1-D pseudo-Wigner distribution (PWD) transformation to the source images and on the use of a pixel-wise cloud model. Both features could also be interpreted as a denoising method centered in a pixel-level measure. Such procedure is able to process sequences of multi-temporal registered images affected with spatial-variant noise. The goal consists in providing a 2-D clean image, after removing the spatial-variant noise disturbing the set of multi-temporal registered source images. This is achieved by taking as reference a statistically parameterized model of a cloud prototype. Using this model, a pixel-wise measure of the noise degree of the source images can be calculated through their PWDs. This denoising procedure enables to choose the noise-free pixels from the set of given source images. The applicability of the method to the cloud removal paradigm is illustrated with different sets of artificial and natural cloudy or foggy images, partially occluded by clouds in different regions. Another advantage of the present approach is its reduced computational cost, once the 1-D case has been preferred instead of a full 2-D implementation of the PWD.