Apache airavata

Clone this repo:

Branches

  1. ace1dc1 Adding repository support for SFTP Remotes by Dimuthu Wannipurage · 8 weeks ago master
  2. 5c072e3 Initial framework for orchestrator API by Dimuthu Wannipurage · 9 weeks ago
  3. d760234 Showing the latest build status by Dimuthu Wannipurage · 3 months ago
  4. 1026697 Enabling the travis build by Dimuthu Wannipurage · 3 months ago
  5. 1eb8a81 Initial project structure for data orchestrator by Dimuthu Wannipurage · 3 months ago

Apache Airavata Data Lake

License Build Status

Apache Airavata use cases enable capture of data from observational and experimental instruments and computations resulting from Airavata's orchestration capabilities. As the data deluges into vast amounts, harvesting the data, capturing metadata, presenting it for easy and controlled access becomes unmanageable.

Airavata data lake will bundle stand alone services to catalog data in various data stores, extract and capture semantics and metadata descriptions of the data and preserve associated data provenance. The data lake will provide API's, query and search capabilities to programmatically search and retrieve data and power building user interactivity and data analysis applications on top of it.

Airavata Data Lake Overview

Airavata Data Lake will provide file watcher and other trigger capabilities to ingest data from scientific instruments as they become available. The framwork will enable pluggable data parsers to read structured and unstructured data files and extract meaningful descriptions.

A bundled Data replica catalogs will associate pointers to data at multiple storgae locations. The replica catalog maps logical file names to the physical locations. Data Lake client SDK's will provide API functions to query replica location and resolve into multiple physical file locations. The client will be bundled with access protocols to retrive the data or to embedd into computational pipelines.

Interfacing with Airavata Managed File Transfer Service Data can moved and archiving into longer term persistant storages like tapped archives. The Data archives will be indexed and have search capabilities

Data Lake's provenance will provide information to capture parameters influenced the production or modification of the data. An abstraction API will enable plugging fine granted provenance based on Airavata tentant context. Interfacing with Airavata Orchestration Services, pointers to experiment catalog will enable restructuring of the underting computations.