SODALITE: Software Defined AppLication Infrastructures managemenT and Engineering
Event Type
Project Poster
TimeTuesday, June 23rd4:32pm - 4:35pm
LocationAnalog 2
DescriptionSoftware application developers and users are now targeting a wide range of diverse computing platforms: clusters and supercomputers with homogeneous or heterogeneous node architectures for heavy batch computations, including resources available on the Cloud. HPC jobs with requirement for specialised execution environments such as compute accelerators or specialised hardware for data movement, are becoming more and more popular. However, the support of specific ICT software technology to manage the heterogeneous infrastructures in compute centres requires standard that are not properly defined. SODALITE (, an Horizon 2020 project proposes to tackle the complexity of deploying and operating modern applications onto heterogeneous HPC and cloud-based systems by providing application developers and infrastructure operators with tools to abstract their application and infrastructure requirements. This enables simpler and faster development, deployment, operation, and execution of heterogeneous applications reflecting diverse circumstances over heterogeneous high-performance and cloud infrastructures, with a particular focus on performance, quality, manageability, and reliability. SODALITE attempts to bring the vast knowledge of performance optimisation accrued by the HPC industry over decades into the cloud computing area. Using application and infrastructure Performance Model, SODALITE deployment framework will enable automated performance optimisation before deployment (static) and also at runtime (dynamic). For static optimisations, 3 broad application types are supported: AI training/Inference, Big Data Analytics and Traditional HPC. Using SODALITE IDE, application experts can select optimisations for these applications, which resolve to native instantiations of the application tuned for execution on the given hardware. Additionally, application runtime parameters can also be auto-tuned, further improving performance.
Poster PDF