Enabling BOINC in cloud Services: CPDN as example

Climateprediction.net (CPDN) is a distributed initiative that aims to run thousands of different climate modeling simulations in order to research the uncertainties on parameters, which is needed to be able to understand how small variations in initital values can affect both models and results of simulated climate. CPDN has been running for more than 10 years and nowadays it is facing a set of new different challenges: • Growing and variable need for new resources, caused by the execution and outputs of new versions of the model (HadGEM) and the processing of huge amount or results that are hard to manage with the current infrastructure that runs over the BOINC framework. • Lack, and need, of more control over the workflow and its intermediate stages. • Streamlining of the costs and budget. This is of special interest for research on-demand projects increasingly requested by external research collaborators and stakeholders.

Palabras clave: Climate prediction, cloud computing, volunteer computer, BOINC