Dynamic Development Support for Highly Concurrent Programs in the Ohua Data Flow Engine

Research output: Contribution to journalConference articleContributed

Contributors

Abstract

Most programs that require the full capability of multi-core architectures in order to achieve scalability address very challenging tasks. In the era of cloud computing, web technologies and big data these programs are often required to be online 24/7. Nevertheless, these programs need to be enhanced with new features or might have bugs to be fixed. Hence, a runtime system is required that allows for dynamic development without halting the executing program. While famous scripting languages like Python or JavaScript already provide such a feature, these languages were not designed for highly concurrent programming. Introducing dynamic development into a highly concurrent runtime system is a challenging task that we address in this paper. We present our dataflow-based execution engine, Ohua, as a promising approach to write and execute highly concurrent programs for the future multi-core era. Furthermore, we extend the principles of flow-based programming in order to create a runtime extension framework that, due to the dataflow abstractions, enables an easy incorporation of new runtime features such as dynamic development into the engine.

Details

Original languageEnglish
JournalLaME '13
Publication statusPublished - 1 Jul 2013
Peer-reviewedNo

Conference

Title Languages for the Multicore Era (LaME '13), 2013.
Abbreviated titleLaME'13
Conference number
Duration1 July 2013
Degree of recognitionInternational event
Location
CityMonpellier
CountryFrance

Keywords

Research priority areas of TU Dresden

DFG Classification of Subject Areas according to Review Boards