The acronym "JUNIPER" stands for "Java platform for hIgh PErformance and Real-time large scale data management", which is an EU FP7-ICT project, started in December 2012, aiming to establish a development platform for new-generation data-demanding applications. The project's duration is 36 Months.
Motivation for the Project
The trend towards Cloud technology, the advances of Internet technology, and the proliferation of mobile computing (in the context of “Internet of Things”) are just a few factors contributing to the massive amounts of data collected by means of the modern IT solutions, which is frequently referred in the Science and Technology as a “Big Data” problem. The age of Big Data presents a variety of challenges in how to store, access, handle, stream, process, and analyze massive amounts of structured and unstructured data effectively. In the near term, the major issue with the Big Data oriented applications will be to find new ways to leverage the technology available today to improve the performance and scalability, reduce response time, figure out more efficient ways to deliver services to end users, and meet the growing demand for virtualized (web and cloud) Big Data services. Using High Performance Computing infrastructures is essential in dealing with these issues.
JUNIPER will exploit the power of heterogeneous high performance (supercomputer) and on-demand (cloud) computing and storage infrastructures in order to help Big Data applications provide real-time performance guarantees. Creation of more effective data processing environments, also in terms of the power consumption, and offering real-time guarantees of the data services are tremendous challenges that JUNIPER addresses. Guided by requirements of enterprise Big Data applications, the project will exploit synergies between the major parallelization technologies (such as MPI, MapReduce, COMPSs) and elaborate new paradigms in data-centric parallel processing that will balance flexibility and performance of data processing applications in heterogeneous computing architectures. JUNIPER will deliver a promising platform for tackling today’s and also tomorrow’s Big Data challenges and embrace new opportunities for the Java programming language to reinforce its place as the major technology for development of the effective and scalable solutions for data-intensive problems.
The JUNIPER platform will empower the developers to transcend the limitations of today’s data processing environments by leveraging the full-backed high performance computing technology. The advantages of using JUNIPER will result in spawning much more users and increasing data services’ transaction rate.
A related issue for the JUNIPER platform is how to simplify the development process and minimize the time-to-market costs of data-intensive applications. The key of the JUNIPER technology is that it will allow the application developers predict the characteristics of the data-centric application workflows, including performance, scalability, throughput, etc., on the available hardware resources, and vice versa, will make suggestions for the computing platform required to gain the specified characteristics. Guided by these requirements, JUNIPER will elaborate methodologies for modelling (e.g. based on UML) application work- and data-flow and will develop necessary tools for this.
Apart from the High Performance Computing Center Stuttgart (Germany), the following organizations participate at the project:
- The Open Group (United Kingdom) - coordinator
- Aicas GmbH (Germany)
- University of York (United Kingdom)
- Scuola Superiore Sant’Anna (Italy)
- Softteam (France)
- HMI-Tech GmbH (Germany)