Data-flow Constructs Research Articles

From Lustre to Graphical Models and SCCharts

We introduce a systematic approach for automatically creating a visual diagram, akin to the graphical Safety Critical Application Development Environment (SCADE) model, from a Lustre program. This not only saves tedious manual drawing effort but also enables modeling software to automatically provide the developer with different meaningful views for the same program. We also extend the Sequentially Constructive Charts (SCCharts) language with data-flow constructs that adhere to the Lustre semantics, which permits a translation from Lustre to graphical SCCharts. This allows using the SCCharts code generation, simulation, and visualization tooling also for Lustre programs, in addition to the already existing Lustre compilation techniques. Furthermore, we investigate how the sequentially constructive model of computation, used in SCCharts and other synchronous languages, can be used to conservatively extend Lustre. We have implemented and validated this work with the Eclipse-based open-source Kiel Integrated Environment for Layout Eclipse Rich Client (KIELER) framework.

Parallel Execution of Data-intensive Web Services Based on Data-flow Constructs and I/O Operation Ratio

Parallel Execution of Data-intensive Web Services Based on Data-flow Constructs and I/O Operation Ratio

Context-aware scientific workflow systems using KEPLER

Data intensive scientific workflows are often modelled using a dataflow-oriented model. The simplicity of a dataflow model facilitates intuitive workflow design, analysis and optimisation. However, some amount of control flow modelling is often necessary for engineering fault tolerant, robust and adaptive workflows. In scientific domain, myriads of environment information are needed for controlling different stages of execution. Modelling the control flow using inherent dataflow constructs will quickly result in a workflow that is hard to comprehend, reuse and maintain. In this paper, we propose a context-aware architecture for scientific workflows. By incorporating contexts within a dataflow-oriented scientific workflow system, we enable the development of context-aware scientific workflows without the need to use numerous low level control flow actors. This approach results in a workflow that is aware of its environment during execution with minimal user input and that responds intelligently based on such awareness at runtime. A further advantage of our approach is that the defined contexts can be reused and shared across other workflows. We demonstrate our approach with two prototype implementations of context-aware actors in KEPLER.

A threaded/flow approach to reconfigurable distributed systems and service primitives architectures

This paper discusses a methodology for managing the assembly, control, and disassembly of large numbers of independent small-scale configurations within large-scale reconfigurable distributed systems. The approach is targeted at service primitives architectures for enhanced telecommunications networks, but can apply to more general settings such as multi-tasking supercomputers and network operations systems. * Study of the methods presented here was a key motivation in founding the Bell Communications Research Integrated Media Architecture Laboratory (IMAL) [1] . The Threaded/Flow approach uses data-flow constructs to assemble higher level functions from other distributed functions and resources with arbitrary degrees of decentralization. Equivalence between algorithms and hard and virtual resources is accomplished via threaded-interpretive constructs. Function autonomy, concurrency, conditional branching, pipelining, and setup/execution interaction are implicitly supported. Some elementary performance comparisons are argued. This work is motivated by telecommunications applications involving coordinated multiple-media in open architectures supporting large numbers of users and outside service vendors. In such networks it is desired that services may be flexibly constructed by the network, service vendors, or by users themselves from any meaningful combination of elementary primitives and previously defined services. Reliability, billing, call progress, real-time user control, and network management functions must be explicitly supported. These needs are handled with apparent high performance by the approach.

A framework for data-flow distributed processing

A framework for data-flow distributed processing is established through the definition of a data-flow model and a set of language constructs for concurrent programming. The proposed approach is based on the following characteristics: i) the exploitation of parallelism at the operation level leads to the efficient and natural exploitation of parallelism at the program level, and ii) parallelism, communication, nondeterminism and history sensitivity are primitive concepts. The aim of the defined data-flow constructs is to enhance modularity and parallelism of programs. Two structuring levels are introduced, called «modules» and «frames», to permit both symmetric and asymmetric communication. Single assignment and guarded commands are employed inside modules. Examples of tipical programming problems, including shared resources management, are given together with a short account of a distributed data-flow architecture able to support data-flow distributed processing efficiently.