Cloud computing promises several advantages over classic IT models and has undoubtedly been one of the most hyped topics in the industry over the last couple of years.
Besides the established delivery models Infrastructure as a Service (IaaS) and Software as a Service (SaaS), especially Platform as a Service (PaaS) has attracted significant attention these days.
PaaS facilitates the hosting of scalable applications in the cloud by providing managed and highly automated application environments.
Although most offerings are conceptually comparable to each other, the interfaces for application deployment and management vary greatly between vendors.
Despite providing similar functionalities, technically different workflows and commands provoke vendor lock-in and hinder portability as well as interoperability.
In this study, we present the tool Nucleus, which realizes a unified interface for application deployment and management among cloud platforms.
With its help, we aim to increase the portability of PaaS applications and thus help to avoid critical vendor lock-in effects.
Modern software systems are becoming increasingly integrated and are required to operate over organizational boundaries through networks. The development of such distributed software systems has been shaped by the orthogonal trends of service-orientation and process-awareness. These trends put an emphasis on technological neutrality, loose coupling, independence from the execution platform, and location transparency. Execution platforms supporting these trends provide context and cross-cutting functionality to applications and are referred to as engines.
Applications and engines interface via language standards. The engine implements a standard. If an application is implemented in conformance to this standard, it can be executed on the engine. A primary motivation for the usage of standards is the portability of applications. Portability, the ability to move software among different execution platforms without the necessity for full or partial reengineering, protects from vendor lock-in and enables application migration to newer engines.
The arrival of cloud computing has made it easy to provision new and scalable execution platforms. To enable easy platform changes, existing international standards for implementing service-oriented and process-aware software name the portability of standardized artifacts as an important goal. Moreover, they provide platform-independent serialization formats that enable the portable implementation of applications. Nevertheless, practice shows that service-oriented and process-aware applications today are limited with respect to their portability. The reason for this is that engines rarely implement a complete standard, but leave out parts or differ in the interpretation of the standard. As a consequence, even applications that claim to be portable by conforming to a standard might not be so.
This thesis contributes to the development of portable service-oriented and process-aware software in two ways: Firstly, it provides evidence for the existence of portability issues and the insufficiency of standards for guaranteeing software portability. Secondly, it derives and validates a novel measurement framework for quantifying portability. We present a methodology for benchmarking the conformance of engines to a language standard and implement it in a fully automated benchmarking tool. Several test suites of conformance tests for two different languages, the Web Services Business Process Execution Language 2.0 and the Business Process Model and Notation 2.0, allow to uncover a variety of standard conformance issues in existing engines. This provides evidence that the standard-based portability of applications is a real issue. Based on these results, this thesis derives a measurement framework for portability. The framework is aligned to the ISO/IEC Systems and software Quality Requirements and Evaluation method, the recent revision of the renowned ISO/IEC software quality model and measurement methodology. This quality model separates the software quality characteristic of portability into the subcharacteristics of installability, adaptability, and replaceability. Each of these characteristics forms one part of the measurement framework. This thesis targets each characteristic with a separate analysis, metrics derivation, evaluation, and validation. We discuss existing metrics from the body of literature and derive new extensions speciffically tailored to the evaluation of service-oriented and process-aware software. Proposed metrics are defined formally and validated theoretically using an informal and a formal validation framework. Furthermore, the computation of the metrics has been prototypically implemented. This implementation is used to evaluate metrics performance in experiments based on large scale software libraries obtained from public open source software repositories.
In summary, this thesis provides evidence that contemporary standards and their implementations are not sufficient for enabling the portability of process-aware and service-oriented applications. Furthermore, it proposes, validates, and practically evaluates a framework for measuring portability.
In 2007, OASIS finalized their Business Process Execution Language 2.0 (BPEL) specification which defines an XML-based language for orchestrations of Web Services. As the validation of BPEL processes against the official BPEL XML schema leaves room for a plethora of static errors, the specification contains 94 static analysis rules to cover all static errors. According to the specification, any violations of these rules are to be checked by a standard conformant engine at deployment time. When a violation is not detected in BPEL processes during deployment, such errors are only detectable at runtime, making them expensive to find and fix.
Due to the large amount of rules, we have created a tag system to categorize them, allowing easier reasoning about these rules.
Next, we formalized the static rules and derived test cases based on these formalizations with the aim to evaluate the degree of support for static analysis of BPEL engines.
Hence, this work is the foundation of the static analysis capabilities of BPEL engines.
Over the last years, the utilization of cloud resources has been steadily rising and an increasing number of enterprises are moving applications to the cloud. A leading trend is the adoption of Platform as a Service to support rapid application deployment. By providing a managed environment, cloud platforms take away a lot of complex configuration effort required to build scalable applications. However, application migrations to and between clouds cost development effort and open up new risks of vendor lock-in. This is problematic because frequent migrations may be necessary in the dynamic and fast changing cloud market. So far, the effort of application migration in PaaS environments and typical issues experienced in this task are hardly understood. To improve this situation, we present a cloud-to-cloud migration of a real-world application to seven representative cloud platforms. In this case study, we analyze the feasibility of the migrations in terms of portability and the effort of the migrations. We present a Docker-based deployment system that provides the ability of isolated and reproducible measurements of deployments to platform vendors, thus enabling the comparison of platforms for a particular application. Using this system, the study identifies key problems during migrations and quantifies these differences by distinctive metrics.
Today, process-aware systems are ubiquitous. They are built by leveraging process languages for both business and implementation perspectives. In the typical context of a Web Services-based Service-oriented Architecture, the obvious choice to implement service orchestrations is still the Business Process Execution Language (BPEL). For BPEL, a variety of open source and commercial engines have emerged. Although the BPEL standard document defines a set of static analysis rules which should be checked by engines prior to deployment to be standard conformant, previous work revealed that most engines are not capable of revealing all violations of these constraints, resulting in costly runtime errors later on. In this paper, we aim to improve the static analysis conformance of BPEL engines. We implement the tool BPELlint that validates 71 static analysis rules of the BPEL specification, show that the tool can be easily integrated into the deployment process of existing engines, and evaluate its performance to measure the effect on the time to deploy. The results demonstrate that BPELlint can improve the static analysis conformance of BPEL engines with an acceptable performance overhead.
Service-oriented systems are increasingly implemented in a process-based fashion. Multiple languages for building process-based systems are available today, but the Business Process Model and Notation (BPMN) is becoming ubiquitous. With BPMN 2.0 released in 2011, execution semantics were introduced, supporting the definition of executable processes. Nowadays, more and more process engines directly support the execution of BPMN processes. However, the BPMN specification is lengthy and complex. As there are no official tests and no certification authority, it is very likely that engines a) implement only a subset of the language features and b) implement language features differently. In other words, we suspect that engines do not conform to the standard, despite the fact that they claim support for it. This prohibits the porting of processes between different BPMN vendors, which is an acclaimed goal of the language. In this paper, we investigate the standard conformance of open source BPMN engines to provide a clear picture of the current state of the implementation of BPMN. We develop a testing approach that allows us to build fully BPMN-compliant tests and automatically execute these tests on different engines. The results demonstrate that state of-the-art BPMN engines only support a subset of the language. Moreover, they indicate that porting BPMN processes is only feasible when using basic language constructs.
In 2007, OASIS finalized their Business Process
Execution Language 2.0 (BPEL) specification which defines
an XML-based language for building orchestrations of Web
Services. As the validation of BPEL processes against the
official BPEL XML schema leaves room for a plethora of static
errors, the specification contains 94 static analysis rules to cover
all static errors. According to the specification, any violations
of these rules are to be checked by a standard conformant
engine at deployment time. When a violation is not detected
in BPEL processes during deployment, such errors remain
unnoticed until runtime, making them expensive to find and fix.
In this work, we investigate whether mature BPEL engines that
claimed standard conformance implement these static rules.
To answer this question, we formalize the static rules and
derive test cases based on these formalizations to evaluate
the degree of support for static analysis of six open source
BPEL engines using the BPEL Engine Test System (betsy). In
addition, we propose a method to get more accurate static
analysis conformance results by taking the feature conformance
of engines into account to exclude false positives in contrast
to the classic approach. The results reveal that support for
static analysis in these engines varies greatly, ranging from
nonexistent to full support. Furthermore, our proposed method
outperforms the classic one in terms of accuracy.
The selection of the best fitting process engine for
a specific project requires the evaluation of engines according
to various requirements. We focus on the non-functional
requirement robustness, which is critical in production environments
but hard to determine. Thus, we propose an evaluation
framework to reveal important robustness criteria of process
engines. In this work, we focus on message robustness, i.e., the
ability to handle the receipt of invalid messages appropriately.
In a case study comprising five open source BPEL engines, we
determine message robustness by injecting faults into robustly
designed processes as a reply to a previously sent request from
an external virtual service and assert their behavior. The results
show that the degree of message robustness significantly differs,
hence, robustly designed processes do not necessarily lead to
robust runtime behavior, the selected engines still play a major
The Web Services Business Process Execution language (BPEL) is a standard
for modeling and executing automated processes and is tailor-made for service
orchestration. BPEL specifies a serialization format which every BPEL implementation
has to understand, thus allowing for the portability of processes among runtime engines.
Although the modeling and execution of BPEL processes is portable between engines
to a large degree, the lifecycle management of BPEL processes is not standardized and
varies a lot for different engines. This paper presents a first approach for a uniform
and cloud-based lifecycle management of BPEL processes and engines. We infer a
uniform interface for the lifecycle management from the capabilities of current engines
and provide a prototypic implementation of a tool that manages processes and engines
on a TOSCA-compliant infrastructure.
Nowadays, business processes and their execution are corner stones in
modern IT landscapes, as multiple process languages and corresponding engines
for these languages have emerged. In practice, it is not feasible to select the best
fitting engine, as engine capabilities are mostly hidden in the engine implementation
and a comparison is hampered by the large differences and high adoption
costs of the engines. We aim to overcome these problems by a) introducing an
abstract layer to access the functionality of the engines uniformly, b) by revealing
the engine capabilities through automated and isolated tests for typical requirements,
and c) support the user in their selection of a process engine by determining
and explaining the fitness of the engines for a single process or a given set of
processes using policy matching against previously revealed engine capabilities.
Early results show the general feasibility of our approach for BPEL engines for a
Keywords: BPM, process engines, engine selection, execution requirements,