Karolina Vukojevic-Haupt

On-demand Provisioning of Infrastructure, Middleware and Services for Simulation Workflows

Service orientation is a mainstream paradigm in business applications and gains even greater acceptance in the very active field of eScience. In SOC service binding strategies have been defined to specify the point in time a service can be discovered and selected for use, namely static binding, dynamic binding at deployment or at run time, and dynamic service deployment. The basic assumption in all these strategies is that the software stack and infrastructure necessary to execute the services are already available. While in service-based business applications this is typically a valid assumption in scientific applications it is often not the case. Therefore, in this work we introduce a new binding strategy for services we call on-demand provisioning which entails provisioning of the software stack necessary for the service and subsequent dynamic deployment of the service itself. Towards this goal, we also contribute a middleware architecture that enables the provisioning of the software stack - functionality unavailable in conventional service middlewares. We demonstrate the approach and the capabilities of the middleware and the current state of the implementation of our approach. For this purpose we use an example application from the field of eScience that comprises a scientific workflow management system for simulations.

Performance and Cost Evaluation for the Migration of a Scientific Workflow Infrastructure to the Cloud

The success of the Cloud computing paradigm, together with the increase of Cloud providers and optimized Infrastructure-as-a-Service (IaaS) offerings have contributed to a raise in the number of research and industry communities that are strong supporters of migrating and running their applications in the Cloud. Focusing on eScience simulation-based applications, scientific workflows have been widely adopted in the last years, and the scientific workflow management systems have become strong candidates for being migrated to the Cloud. In this research work we aim at empirically evaluating multiple Cloud providers and their corresponding optimized and nonoptimized IaaS offerings with respect to their offered performance, and its impact on the incurred monetary costs when migrating and executing a workflow-based simulation environment. The experiments show significant performance improvements and reduced monetary costs when executing the simulation environment in off-premise Clouds.

Service Selection for On-Demand Provisioned Services

Service selection is an important concept in service oriented architectures that enables the dynamic binding of services based on functional and non-functional requirements. The introduction of the concept of on-demand provisioned services significantly changes the nature of services and as a consequence the traditional service selection process does not fit anymore. Existing approaches for service selection rely on the always on semantic of services, an assumption that is not valid for on-demand provisioned services. We tackle this problem by adapting the traditional service selection process and by defining an additional step covering the changes introduced by the concept of on-demand provisioning. Our solution comprises an extended architecture for on-demand provisioning, a metamodel for a service registry, and a detailed definition and discussion of the adapted and extended service selection process. The work presented in this paper allows keeping the advantages of dynamic service binding at runtime and combining them with the advantages of Cloud computing exploited through the concept of on-demand provisioning.

Decision Support for the Migration of the Application Database Layer to the Cloud

Migrating an existing application to the Cloud is a complex and multi-dimensional problem requiring in many cases adapting the application in significant ways. Looking specifically into the database layer of the application, i.e. the aspect providing data persistence and manipulation capabilities, this involves dealing with differences in the granularity of interactions, refactoring of the application to cope with remote data sources, and addressing data confidentiality concerns. Toward this goal, in this work we present an application migration methodology which incorporates these aspects, and a decision support, application refactoring and data migration tool that assists application developers in realizing this methodology. For purposes of evaluating our proposal we present the results of a case study conducted in the context of an eScience project.

Bootstrapping Complex Workflow Middleware Systems into the Cloud

The use of Cloud infrastructures together with provisioning technologies can be successfully applied in scenarios where resources are only needed rarely and irregularly, for example simulation workflows in the eScience domain. There has already been proposed a solution for the on-demand provisioning of services required by workflows, but how to automatically provision the needed workflow middleware itself is still an open issue. Although many provisioning technologies are available, it is currently not possible to use them in an integrated, flexible and automated way. The main idea presented in this paper is a multi-step bootstrapping process, starting with a minimal local software component and ending up with a complex workflow middleware running in the Cloud. This minimal software component is called bootware. We define the key requirements for the bootware, present its architecture and discuss the main design decisions and how they fulfil the requirements. The bootware enables to provision complex workflow middleware systems on-demand and automatically in the Cloud and therefore reduces resource consumption and costs.

Service Composition for REST

One of the key strengths of service oriented architectures, the concept of service composition to reuse and combine existing services in order to achieve new and superior functionality, promises similar advantages when applied to resources oriented architectures. The challenge in this context is how to realize service composition in compliance with the constraints defined by the REST architectural style and how to realize it in a way that it can be integrated to and benefit from existing service composition solutions. Existing approaches to REST service composition are mostly bound to the HTTP protocol and often lack a systematic methodology and a mature and standards based realization approach. In our work, we follow a comprehensible methodology by deriving the key requirements for REST service composition directly from the REST constraints and then mapping these requirements to a standard compliant extension of the BPEL composition language. We performed a general requirements analysis for REST service composition, defined a meta model for a corresponding BPEL extension, realized this extension prototypically and validated it based on a real world use case from the eScience domain. Our work provides a general methodology to enable REST service composition as well as a realization approach that enables the combined composition of WSDL and REST services in a mature and robust way.

A Framework for the Structural Analysis of REST APIs

Today, REST APIs have established as a means for realizing distributed systems and are supposed to gain even more importance in the context of Cloud Computing, Internet of Things, and Microservices. Nevertheless, many existing REST APIs are known to be not well-designed, resulting in the absence of desirable quality attributes that truly RESTful systems entail. Although existing analysis show, that many REST APIs are not fully REST compliant, it is still an open issue how to improve this deficit and where to start. In this work, we introduce a framework for the structural analysis of REST APIs based on their description documents, as this allows for a comprehensive, well-structured analysis approach that also includes analyzing the corresponding API description languages. A first validation builds on a set of 286 real world API descriptions available as Swagger documents, and comprises their transformation into a canonical metamodel for REST APIs as well as a metrics-based analysis and discussion of their structural characteristics with respect to compliance with the REST architectural style.

On-demand provisioning of workflow middleware and services into the cloud: an overview

One of the core principles in service oriented computing is that services are always on and available. There are however domains where running services all the time is not suitable, for example when applying simulation workflows in the eScience domain. The simulation services orchestrated by these workflows are typically used only rarely and irregularly, keeping them running all the time would result in a significant waste of resources. As a consequence, we developed the approach of on-demand provisioning of workflow middleware and services. In this paper we will give an overview about our work. We will present the motivation and main idea of our solution approach and will also provide details about some of the results of our work. The overview about our previous and current work is then complemented by a detailed discussion and comparison of the roles involved in both concepts, traditional service oriented computing as well as our newly developed on-demand provisioning approach.

Performance and Cost Trade-Off in IaaS Environments: A Scientific Workflow Simulation Environment Case Study

The adoption of the workflow technology in the eScience domain has contributed to the increase of simulation-based applications orchestrating different services in a flexible and error-free manner. The nature of the provisioning and execution of such simulations makes them potential candidates to be migrated and executed in Cloud environments. The wide availability of Infrastructure-as-a-Service (IaaS) Cloud offerings and service providers has contributed to a raise in the number of supporters of partially or completely migrating and running their scientific experiments in the Cloud. Focusing on Scientific Workflow-based Simulation Environments (SWfSE) applications and their corresponding underlying runtime support, in this research work we aim at empirically analyzing and evaluating the impact of migrating such an environment to multiple IaaS infrastructures. More specifically, we focus on the investigation of multiple Cloud providers and their corresponding optimized and non-optimized IaaS offerings with respect to their offered performance, and its impact on the incurred monetary costs when migrating and executing a SWfSE. The experiments show significant performance improvements and reduced monetary costs when executing the simulation environment in off-premise Clouds.

A Middleware-Centric Optimization Approach for the Automated Provisioning of Services in the Cloud

The on-demand provisioning of services, a cloud-based extension for traditional service-oriented architectures, improves the handling of services in usage scenarios where they are only used rarely and irregularly. However, the standard process of service provisioning and de-provisioning shows still some shortcomings when applying it in real world. In this paper, we introduce a middleware-centric optimization approach that can be integrated in the existing on-demand provisioning middleware in a loosely coupled manner, changing the standard provisioning and de-provisioning behavior in order to improve it with respect to cost and time. We define and implement a set of optimization strategies, evaluate them based on a real world use case from the eScience domain and provide qualitative as well as quantitative decision support for effectively selecting and parametrizing a suitable strategy. Altogether, our work improves the applicability of the existing on-demand provisioning approach and system in real world, including guidance for selecting the suitable optimization strategy for specific use cases.