News / 23 September 2019

Drones, Smart Farms, and iRODS

The iRODS Consortium has recently joined forces with Australia Agriculture Victoria's pilot effort to integrate the benefits of storage virtualization and metadata-driven, open-source data management with their already considerable efforts with agricultural technology.

The ultimate goal is to create an IoT-and-web infrastructure for synthesizing, visualizing, and sharing data gathered from a varied collection of sensors deployed on "Smart Farms". These data could be about soil and crops, air and water quality, or livestock.

As the first part of the effort, the Consortium stood up a single-server iRODS Zone capable of registering a drone-gathered data set stored in an Amazon S3 bucket. Upon ingest, the server then extracted key features from the embedded TIFF files and attached them as metadata to the newly registered data objects within iRODS.

Powered by the iRODS automated ingest tool and cacheless S3 storage plugin, the initial testing showed roughly 20 objects registered per second, per core, but reading more than 70% of each data file while processing the EXIF tags from TIFF.

Upon further investigation and iteration, the iRODS Consortium was able to extract the same EXIF data after accessing only 1% of each data file from S3. This proves an economical way to operationalize valuable provenance information about the source data.

Metalnx and Davrods services are deployed on the server too, which will immediately allow collaborators and other interested parties to view the ingested data and metadata using a standard set of graphical Desktop tools, mainly file managers and browsers.

This new collaboration with AgVic has only begun to scratch the surface, as eventually the intent is not only to harness well-known iRODS capabilities in automating computational workflows, but also to make the geo-connected result data easily searchable via iRODS' newly developed indexing framework.

Daniel Moore