Provision OSCAR clusters on on-premises, public and federated Clouds
Kubernetes clusters grow and shrink according to the workload
Compose data-driven serverless workflows with a Functions Definition Language
Built on Kubernetes
OSCAR’s services use Kubernetes components for easier extensibility
Distributed under the Apache 2.0 License in GitHub. Also offered as SaaS
Serverless for Compute-Intensive Processing
OSCAR supports data-driven serverless computing for file-processing applications. Services will be triggered in response to a file upload to an object storage back-end in order to execute a user-defined shell script inside a container provisioned out of an user-defined Docker image. These will be orchestrated as a Kubernetes batch jobs. The output data will be uploaded to any object storage back-ends support. Synchronous invocations available.
Support for Multiple Storage Back-ends
Each OSCAR cluster features a MinIO installation so that file uploads trigger the execution of the file-processing applications. Other storage back-ends are supported for file storage output including Amazon S3 and the EGI DataHub (based on Onedata). These can be chained to create data-driven workflows of functions
An OSCAR cluster is entirely based on dynamically deployed elastic Kubernetes clusters that can grow and shrink in terms of the number of nodes thanks to the CLUES elasticity system. Clusters self-adapt to the incoming workload by provisioning additional nodes up the limit configured at deployment time.
Automated Deployment on Multi-Clouds
An OSCAR cluster can be provisioned from the CLI using EC3 but the simplest approach is to use the Infrastructure Manager (IM) Dashboard, which provides a streamlined process of deploying the cluster on any Cloud that you have access to.
Serverless Workflows for the Cloud Computing Continuum
OSCAR is integrated with SCAR, an open-source tool to execute generic applications on AWS Lambda, the Functions as a Service (FaaS) provided of Amazon Web Services (AWS). This allows to create serverless workflows across the Cloud computing continuum, where some lightweight processing occurs in an on-premises Cloud (or in the edge) and intensive computing takes place in AWS Lambda. SCAR is also integrated in with AWS Batch, a managed service to provision auto-scaled clusters in AWS Batch. This allows to execute event-driven serverless workflows for applications that require intensive computing or specialized accelerated hardware such as GPUs.