Generation of feasible deployment configuration alternatives for Data Distribution Service based systems

Bedir Tekinerdogan, Turgay Çelik, Ömer Köksal*

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

3 Citations (Scopus)

Abstract

Data distribution service (DDS) has been defined by the OMG to provide a standard data-centric publish-subscribe programming model and specification for distributed systems. DDS has been applied for the development of high performance distributed systems such as in the defense, finance, automotive, and simulation domains. To support the analysis and design of a DDS-based distributed system, the OMG has proposed the DDS UML Profile. A DDS-based system usually consists of multiple participant applications each of which has different responsibilities in the system. These participants can be allocated in different ways to the available resources, which leads to different configuration alternatives. Usually, each configuration alternative will perform differently with respect to the execution and communication cost of the overall system. In general, the deployment configuration is selected manually based on expert knowledge. This approach is suitable for small to medium scale applications but for larger applications this is not tractable. In this paper, we provide a systematic approach for deriving feasible deployment alternatives based on the application design and the available physical resources. The application design includes the design for DDS topics, publishers and subscribers. For supporting the application design, we propose a DDS UML profile. Based on the application design and the physical resources, the feasible deployment alternatives can be algorithmically derived and automatically generated using the developed tools. We illustrate the approach for deriving feasible deployment alternatives of smart city parking system.
Original languageEnglish
Pages (from-to)126-145
JournalComputer Standards and Interfaces
Volume58
Early online date10 Jan 2018
DOIs
Publication statusPublished - May 2018

Keywords

  • Data Distribution Service (DDS)
  • Design optimization
  • Feasible deployment
  • Model-driven development
  • Software architecture analysis

Fingerprint Dive into the research topics of 'Generation of feasible deployment configuration alternatives for Data Distribution Service based systems'. Together they form a unique fingerprint.

  • Cite this