Accelerator

IIASA Accelerator

The IIASA Accelerator is a cyberinfrastructure conceptualized to efficiently integrate models and data, explore model outputs and disseminate outcomes linked to iBIOM.

Scalable data store and scalable computing capacity are essential when dealing with big data and big data analysis, as in the case of the iBIOM framework. The Accelerator features a DFS for data scalability and a high-throughput computer (HTC) cluster for computing power. A platform control server stays at the top of this architecture and acts as a controller for the DFS and HTC cluster. Requests to the platform control server are made from a platform client application and are usually responsible for exploring scenarios or simulations results and dispatching workflows related to simulations to the HTC.

Jobs, including models and data processing scripts, are expected to be containerized and saved as a container image on a container image registry. When the jobs are dispatched, the HTC will pull the job container image from the registry and execute it, while all the input/output operations are performed on the DFS or other remote storage systems.

To handle spatially explicit datasets, such as maps and gridded modelling results, a geo server can directly read the data from the DFS and serve it to GUI clients efficiently.