Migrating industrial automation systems to a public cloud would deliver
several advantages such as cost reduction for system developers and users.To close the availability gap,we propose a reliable cloud-based feedback control system.The proposed system achieves high availability of feedback controllers by using multiple regions and peer- to-peer communication for controller status sharing.The developed system shows that the feedback controllers can perform failover for two types of failures,namely,network failure and controller failure. Figure 1 depicts a conventional industrial automation system.The system manages states of devices such as airconditioning systems or room lights in a building. The system consists of several controllers that monitor and control devices precisely. In this paper,as in conventional works ,we focus on the availability of the controllers since there are plenty of works on typical server applications such as HMI and DB. Several methods that aim to achieve high availability of cloud-based controllers have been proposed recently .Those methods run multiple controllers on a cloud and select one of them as a primary controller by using a gateway or a server that manages controllers’ information required for primary selection. Although,like conventional methods,the proposed system runs multiple controllers on multiple regions of a cloud,the controllers in the proposed system exchange heartbeat messages to detect a failure of other controllers,and therefore the controllers can perform failover by themselves. The following subsections explain the detail of the controllers. Reliable cloud-based feedback control system that of LAN. Each controller performs PID feedback control with the system,the controllers share some information via the gateway placed in a building to detect a failure of the primary controller. The system depends on the GW because the controllers cannot determine a primary’s failure without the GW.Although the system we propose is similar to the system proposed in ,it does not need the GW for failover. Feedback control diagram of the controller The proposed control system is available only if one of the regions and one of the access networks and the internet are available.The availability of the proposed system Ps is formulated as. In the proposed system,the controllers can perform failover without dependency on any other gateways or servers. Engine,guarantee availability for each region instead of each in our proposed system,the controllers send Heart Beat messages periodically to one another to inform their statuses to other controllers.Each controller checks the status of its control network,a network between a controller and the GW,and puts the status into HB to share with other controllers.The controller judges the status of its control network is normal if it gets a device state correctly.The primary controller starts the primary selection procedure when it notices the failure status. The delay can be measured when the controller gets a device state if the messaging pattern of the communication protocol is a request/response pattern .Calculates a control value if the controller is a primary.Then the primary sends the value to the GW.A controller sends HB messages to other controllers . If the sender is the primary and the sender’s control network status is failure,the receiver starts the primary selection procedure.The sender starts the primary selection procedure if the receiver is the primary and it is in failure status. Destination min ave max stddev no failure no sharing sharing PID and Smith Predictor. Each controller starts new primary selection if it detected a failure of the primary controller or a failure of the primary’s control network.The controllers whose control network status is normal are the candidates for the new primary.The new primary should be decided by considering communication delay since it is an important factor for control quality.Each controller calculates the score Sprimary for each candidate and determines one of the candidates that has minimum score as the new primary. We have developed the proposed system as a proof of concept.The controller uses the AMIGO method to tune gain parameters of PID.The controller gets a device state from the GW by using a simple original UDP-based protocol and its sampling interval is 1 second. In this evaluation,we confirm that sharing statuses of the primary controller is deployed in Tokyo region and the backup controller is deployed in NC region.We blocked the communication of the primary’s control network at 10 seconds from the beginning of the evaluation to emulate a failure of the control network.We set Ncn = 3 and the sampling interval is 1 second,and thus the primary controller is expected to detect a failure of its control network at around 13 seconds.We implemented the controller and the GW as Java-based software. The controllers run on Ubuntu instances of Amazon EC2.Figure 4 shows the device states in this evaluation.The vertical line at 14 seconds means the time when the backup controller became a new primary controller.Without control status sharing,the device states show large disturbance after the failover. It would be better if the system could forecast a long-term failure.
Towards Automated Workflow Deployment in the
Cloud using TOSCA This paper shows how TOSCA,a new standard for cloud service management,can be used to systematically specify the components and life cycle management of scientific workflows by mapping the basic elements of a real workflow onto entities specified by TOSCA.Ultimately,this will enable workflow definitions that are portable across clouds,resulting in the greater reusability and reproducibility of workflows. It aims to enable the automated deployment and management of cloud applications.TOSCA is generic enough to cover a variety of scenarios and also portable between different cloud management environments .In this paper we present our work on using TOSCA as a language to describe workflows,workflow components and templates.We want to offer them as reusable entities that include not only the scientific experiment but also all details needed to deploy and execute it. In this paper,to demonstrate our approach in practice,we model an existing workflow using TOSCA.The example involves a typical scientific workflow,i.a set of tasks with data dependencies expressed as a directed acyclic graph.We use TOSCA to represent workflow components and the workflow itself but also to capture the configuration of the whole application. Overall,we show how to utilise the standard to generate a TOSCA-compliant topology for scientific workflows.TOSCA.They require multiple components to be deployed and configured before and during runtime.For a scientific method to be effectively reused over time,and for experiments to reproduced,the repeatability of these deployment and configuration steps is crucial. Otherwise,the value of building workflows is quickly lost .Unfortunately,it is impractical to expect most scientists to perform these complex deployment steps manually.TOSCA is a specification for modeling a complete application stack,and automating its deployment and management in the cloud .The specification defines a meta-model for describing both the structure and management of IT services. The structure of a service is represented by the Topology Template which consists of Node and Relationship Templates.Together they define a service as a directed graph of deployable components.Together they are able to describe the logical relationships and other dependencies between the application’s node templates .The deployment process. Plans encode a sequence of operations required to instantiate TOSCA services and thus they follow an «imperative» approach.Often,a TOSCA runtime environment is able to infer a correct deployment plan and management procedure only by interpreting service topology.The main advantage of the declarative approach is that it hides low-level deployment activities from the user.Scientists can focus on the definition of the high-level architecture of their experiment,which the TOSCA runtime can translate into a detailed deployment procedure. In this work we therefore adopt the declarative approach and use the Topology Template to define workflows.As mentioned earlier,TOSCA is still an emerging standard.At the time of writing this paper,the YAML-based version of the specification has not yet been released,thus we use vendorspecific flavour of TOSCA YAML provided by Cloudify.Cloudify is a free and open-source orchestrator platform that intends to use TOSCA to automate the deployment and scaling of applications over any cloud technology. In this section we show how TOSCA can be used to specify a scientific workflow,including discussion of the different stages followed to create a complete service template.The first step to model a workflow using TOSCA is to identify all its constituent parts.These include workflow tasks and all their software dependencies such as the specific packages and libraries required by the tasks to run.Node types are usually derived from the basic types provided by TOSCA and Cloudify DSL,like ApplicationModule,and then customised with specific property and interface definitions. These types define an interface with operations to configure the source and target nodes joined by the relationship.Among the basic relationship types one of the most common is contained in.It is used to create vertical software stack like virtual machine that hosts an operating system which in turn hosts one or more workflow services.When connecting new,non-standard node types,a new type of relationship may be required. The relationship definition is used to specify the semantics of a link between nodes and also methods which realize such a link.For example,the connected to relationship needs implementation of methods which can bind two end nodes,as in a client-server connection.The TOSCA metamodel uses the concept of a Service Template to describe a cloud application.We use it to model the high-level structure of scientific workflows. Service Template is a graph of Node Templates which represent specific instances of application components and Relationship Templates that model links between these instances.Clearly,it fits the notion of scientific workflow very well. Originally,it was designed in the e- Science Central system The result is a self-contained and portable service model that can be used to deploy and manage workflow instances in the cloud.It includes a cloud-based workflow enactment engine to which users can submit their workflows via a web browser or desktop application.The system implements a simple dataflow model in which workflows are built from blocks connected into a direct acyclic graph.e-SC workflow comprises a comprehensive list of components,services,and assemblies required to achieve specific functionality implemented by workflow blocks. This paper has shown that the TOSCA specification can fulfil this need for scientific workflows.We have presented the first attempt to use TOSCA to formally describe the internal topology of a scientific workflow,together with its deployment processes.
A Reliable Cloud Based Feedback Control System
Section I introduces operational and performance models of a system
of shared resources.Section II analyzes a symmetric system under mean-field approximation and proposes a Perron-Frobenius based approach to analysis of non-symmetric systems under fluid approximation.We argue that economics makes this tradeoff more pronounced by driving Cloud service providers towards the boundary of the operational regime,and thus increasing risk of overload when the system does not have sufficient capacity for sustaining the exogenous demand.Subsection A introduces a performance model of statically shared resources,and demonstrates that under broad assumptions economic pressures drive system towards full utilization. Subsection B introduces a performance model of dynamically shared resources.Keywords-cloud computing model,dynamic resource sharing,overload,systemic risk. These tradeoffs are due to the benefits of accommodating occasional resource demand/supply imbalances being inherently associated with risks of local overload spreading over a sizable portion of the system. Due to intractability of the conventional performance models of a realistic size Cloud,we employ methodology of Complex. As opposed to the conventional view of systemic overload as being continuous with respect to the exogenous utilization,our analysis under mean-field and fluid approximations [3]- [4] indicates a possibility of abrupt/discontinuous systemic overload,which results in the system transitioning to a persistent congested mode through cascades of local overloads. The existence of such a solution indicates a correspond to the globally stable «normal» and «congested» system equilibria respectively.Branches A* A* and possibility of the system transitioning to a persistently congested mode despite long-term resource demand/supply is balanced.This transition occurs through cascades of local congestion spreading to other parts of the system due to dynamic resource sharing.B* B* correspond to the coexisting «normal» and congested metastable system equilibria respectively for intermediate load. Subsection A analyzes symmetric system under mean-field approximation slowly changing level of resource sharing ,indicates a combination of positive and negative effects of the dynamic resource sharing on the system performance. Consider a particular case of symmetric system with native services ,where Figure 1 sketches the persistent loss L vs.Curve 0E0 sketches loss rate L for sufficiently low level of resource sharing ,when mean-field equation has unique for all. Persistent loss vs.level of resource sharing. As resource sharing «slowly» increases ,loss rate. For this paper has suggested that the economic benefits of dynamic resource sharing are inherently associated with systemic risk of overload,which may be either gradual/continuous or abrupt/discontinuous.Of particular interest is a potential ability of online measurements of the corresponding Perron- Frobenius eigenvalue to provide «early warning signals» of the system approaching the instability/breaking point] for the purpose of initiating appropriate control actions. As exogenous slowly increases from 0, system follows 0 ABDE .As exogenous load slowly decreases to 0 ,system follows curve EDCA. discontinuous overload,which is based on type of bifurcation of the fluid approximation ,on the boundary of the operational regime.
Systematic Risk in the Cloud Computing Model:
Complex System Perspective Probably public clouds either offering IaaS or those running today’s Internet are the lowest cost solution and in aggregate are far more powerful than the systems used in science research.Of course all systems require a significant ecosystem with many people developing,testing and running software.Big Data systems.In this vision paper,we purport to examine the relationship between infrastructure for data-intensive computing and that for High Performance Computing and examine possible convergence of capabilities. We have examined extensively the landscape of applications across the HPC and data-intensive spectrum.Ogres provides a classification and structure including,classic MPI-based simulations,pleasingly parallel and workflow systems,and dataintensive applications epitomized by deep learning.We highlight the primary differences and similarities between data-intensive problems and traditional highperformance applications. Traditional exascale simulations involve applications of differential equation based models that need very fine space and time steps and this leads to numerical formulations that need the memory and compute power of an exascale machine to solve individual problems .Big data problems do not clearly need a full exascale system to address a single job.Typically any platform will be running lots of jobs that are sometimes pleasingly parallel/MapReduce and sometimes small to medium size HPC jobs which in aggregate are exascale .In fact many public clouds now offer features characteristic of HPC including GPU’s,high performance networks and FPGA accelerators. This is clear for deep learning in the cloud where the value of GPU’s Is well understood.2017.120 the MapCollective model,and Simulations tend to need high precision and very accurate results ,however,data- intensive problems often don’t need high accuracy as seen in trend to low precision deep learning networks,as there are no derivatives and the data has inevitable errors.There are similarities between graph based data intensive applications and particle simulations with a particular cutoff force.Both are MapPoint-to- Point problem architecture many data-intensive problems involve full matrix algorithms and thus are easy to parallelize similar to «long range force» as all points are linked. Data ,as well as data-intensive applications that will increasingly need HPC .Given the current separation of characteristics of HPC and data- intensive applications this requires a convergence of capabilities. Big Data have similar requirements.This requires the selective integration of the Apache Big-Data Stack capabilities appropriately implemented for supercomputing platforms.Data Stack where we examined the addition of high performance runtime and components to Apache systems.We have highlighted the importance of the Big Data systems associated with Apache Foundation,such as Hbase,Hadoop,Spark,Storm etc. Stack ,even though important components such as We note that most of these technologies are in principle usable on both HPC and Cloud IaaS systems,though in practice many challenges remain.ABDS technologies.
Cloud Computing and Privacy Regulations: An
Exploratory Study On Issues and Implications.
A new paradigm in information technology advancement is cloud
computing. The use and development of cloud computing technology is substantial and comes in many forms. The enterprise's people, processes, and technology are all impacted by cloud computing. Despite the efficiency, flexibility, ease of setup, and overall decrease in IT costs that the cloud computing paradigm offers, there may be privacy and confidentiality hazards. The hazards to privacy and secrecy are not the same for all forms of cloud computing. Some predict that in the future, a large portion of the computing activity currently carried out solely on computers that individuals own and operate locally will move to the cloud. When using cloud computing, users connect to the CLOUD, which, in contrast to conventional computing, appears as a single entity. Cloud computing is a new paradigm in the world of Information Technology Advancement.Considerable amount of cloud computing technology is already being used and developed in various flavors. Cloud Computing affects people, process and technology of the enterprise. Inspite of having benefits with Cloud computing paradigm such as efficiency, flexibility, easy setup and overall reduction in IT cost, cloud computing paradigm could raise privacy and confidentiality risks. “Not all types of cloud computing raise the same privacy and confidentiality risks. Some believe that much of the computing activity occurring today entirely on computers owned and controlled locally by users will shift to the cloud in the future”. In Cloud computing, users connect to the CLOUD, which appears as a single entity as opposed to the traditional way of connecting to multiple servers located on company premises. Public Private Partnership these days is a usually adopted pattern of governance to meet the diverse needs of their citizens with confidence and providing quality of these services. Cloud Computing Technology can also act as a facilitator between public and private partnership. In such cases there is a possibility that an external party can be involved in providing Cloud Services having partial control over the data storage, processing and transmission of data and privacy regulations become relevant. Cloud computing has significant implications for the privacy of personal information as well as for the confidentiality of business and governmental information. A survey by EDU CAUSE involving 372 of its member institutions revealed that a great proportion of the respondents with use cases that involved cloud-based services reported that data privacy risks and data security risks were among their top barriers to overcome. A principal goal of this paper is to identify privacy and confidentiality issue that may be of interest and concern to cloud computing participants and users]. Thus this paper explores to elicit possible issues and regulations in the area of privacy that affect the implementation of Cloud Computing Technologies.