Constellation: a Multi-Cloud Application Designed with the MODAClouds Integrated Modelling Environment

Editor's Notes

• #2: Good morning, I am Antonin Abherve , Software Engineer from Softeam and will present works realized during the last period on the Constellation case study (ex Project Management Server). > First I will present the case Study goals and requirements. > Second, I will talk about how and why we have integrated tools and methods developed in MODAclouds project to our case study. > And finally, I will present the future of Constellation server.
• #3: Let’ start by a short presentation of the context in which the case study has been developed. Softeam is a French IT consulting company which work essentially on software engineering domain. Softeam is also a software vendor and develop two products: ECities, a workflow tool and MODELIO, an UML/BPM modeling tool. Modelio is a modeling solution delivering functionality for business, software and infrastructure architectures with Code and Documentation generation capabilities. Our case study, the constellation server is a web application which provide new functionalities related to Modelio.
• #4: Now I will present our case Study goals and requirements and why we need to deploy it in a multi-cloud environment. Three years ago, we released a new version of Modelio. This release includes a new approach to organize projects and models in Modelio. From this date, we have define à Modelio project as a composition of several model fragments. A model fragment is a part of model stored in specific kind of repository ,based on specific technology wits specific features. (Writable, Unhittable, Local or Remote, Supporting or not Collaborative Work). Complexifications resulting of this new approach lead us to develop a new server to manage Project configurations, access right management and administration of remote model repository.
• #5: Now i will present a short demo of Constellation and it integration with MODAClouds First, what is Constellation concretely ? Constellation is a web application witch allow us to manage all artifacts witch compose a Modelio project : Domains , Users, Work Fragments, Library, Modules. > Work Fragments are writable model fragments. They can be local or Remote. In Case of remote Fragments, the instances who host this fragments are managed by Constellation. > Libraries are Remote unwritable model fragments. They are accessible using HTTP protocol and hosted on instances managed by Constellation. > Modules are Functional extension for Modelio. Witch all of this element, we compose a project. As Modelio user, if I am register on a project,
• #6: What does we expected of the MODACloud project? > First, a way to support our clients’ heterogeneity. Some of our customers want a dedicated server hosted on private cloud. Some others want to have to have an access to constellation as a Service. The number of Constellation users for one client can ben 5 peoples 3 hundread. Resulting of this situation we need to be able to adapt our deployment architecture quietly and at low cost We will see that Creator 4Cloud and CloudML 4cloud help us to archive this goal > The second point is that we need to be cost effective. It’s means , providing this services at chipper price. iit’s very important wjen we deliver Constellation as Software as a service approach. Because In this case, directs cost are supported by us. We will see that the Decision Support System (Venus 4Cloud) and Costs Evaluation and Performance AnalysisTool (SPACE Dev 4Clouds) help us in this way > Finaly, we need to be adaptive : We didn’t want to have to redefine a new process to deploy and maintain the server in each new configuration, for each new cloud providers. Q And we will see that tools of Runtime platfrom provide supports to archive this goal
• #7: Now moving on why and how we have integrated tools and methods developed in MODAcloud project to our case study. In month twenty-for, we have integrated and evaluated some design time tools : Creator 4Clouds, CloudML 4Clouds , SPACE Dev 4Clouds During the last periods we have continued our work on time Design tools and integrate some runtime tools. At deign time, we have used Venues 4Clouds At runtime we have integrated Model @ runtime , Tower4Cloud, SPACE Ops 4Clouds and Feedback Loop We will see how we have used this tools in the next phase of this presentation
• #10: Coonnect to sw eng perspective
• #12: Update this figure
• #15: Number of supported IaaS >= 3 (>= 1 at M18; >=3 at M30) Amazon EC2, Flexiant, CloudSigma, Openstack, Azure plus all IaaS supported by jCloud. Number of supported PaaS >= 2 CloudML and the Models@Runtime engine support Cloud Foundry (public and private instances), AWS RDS, AWS SQS, AWS Beanstalk. In addition, the CPIM library offers support for various services of Azure, Google App Engine and Amazon. Number of Cloud providers supported by the deployment and monitoring solutions >= 5 Tower 4Clouds monitoring platform was tested on: Flexiant, Amazon EC2, Microsoft Azure, Heroku, OpenNebula, OpenStack, Eucalyptus. The Models@Runtime engine was tested against: EC2, Flexiant, OpenStack, CloudSigma, Azure, Pivotal, Cloud Foundry, CloudBees, AWS RDS, AWS SQS, AWS Beanstalk
• #17: We used Creator4Clouds to : > Define a cloud ready architecture based on an assembly of scalable components. > Next, We used Creator4Clouds to model QoS constraints of our application and Plan monitoring strategy by definding a set of monitoring rules. > Finally, we used CloudML 4Cloud integrated to Creator 4 Cloud to model and test several deployment configurations for our application. As result : > Creator 4 Cloud provide us a methodological approach to develops a scalable multi-cloud application architecture > The CPIM /CPSM approach allow us to model on time our application architecture and to deploy it fooling several deployment configuration in different cloud provides. > This approach allowed us to limit the migration cost form on cloud provider to another cloud provider
• #21: Based on Application architecture, deployment models and QoS constraints defined in Creator 4Cloud, we used SPACE Dev 4Clouds to analyses early several deployment scenarios.   As result of the usage of this tool, we obtain an estimation of expected costs of each analyzed deployment scenarios. With this informations, We were able to chose an initial deployment scenario for our application. This deployment is adapted (in therm of size and number of instances for example) to our QoS constraints
• #22: Layered Queueing Networks Stochastic models well-suited for software applications Extended Queueing Networks Apt to model: Interactions between Components Asynchronous/Synchronous calls Resource pooling Tool support: LQN Solver (LQNS) - U. Carleton LQN EngINE (LINE) – MODAClouds Fluid Solver for Layered Queueing Network Models Efficient and Scalable evaluation Fluid models: Set of ordinary differential equations Numerical solution Goes beyond the mean values studied by traditional techniques, e.g., time-varying metrics, transients, ...
• #30: So what can we conclude from our usage of the tools and methods developed by modaclouds in context of our case study. About the devlopement of an application dedicatred to a cloud computing environment: In Current way, we had to configure , deploy and administrate manually our application in cloud environment,. We have also to adapt this application for each cloud provider With MODAClouds, we model our application architecture once and we gain the ability to deploy it foloing multiple deployment strategy on multiple clouds platforms About quality of Service management In Current way, we have to define a quality management process for each cloud platforms. With modacloud we can master the quality of each platform and integration new constraints related to multi cloud About Cost estimation : In Current way, we have to used Ad-hoc methods for costs and performance estimations With MODACLouds, we have a fine estimation of costs and performance before deploymentl. Finay , About runtimed management : In traditional way, we have to develop specific strategy to manage deployment, monitoring, scaling our components for each cloud providers With MODAClouds, whe have a global way to manage the life cycle of our application.
• #31: One year ago, we released a first commercial version of constellation > This Application has been devloped during the MODAClouds project > During his devlopement, we used MODAClouds tools ( Creator 4Cloud, CloudML 4Cloud, SPACE Dev 4Clouds) > The next update of constellation will integrate MODACLouds runtime component (Tower 4Clouds and SPACE Ops 4Clouds ) With the experience acquire during the project, we are working now on an Software As A Service offer for constellation. And in close future, we plane to develop new Cloud Services in relation with Constellation and Modelio.