filmov
tv
Lessons learned deploying a big-data geospatial system on AWS

Показать описание
by Eric Wenger
During the past year we undertook an effort to transition a geospatial data processing system developed under a government research project from a small private cluster running Windows to an AWS-based environment running Linux. This talk will cover a variety of subjects including the steps taken to migrate the system, hurdles encountered with Linux and AWS (particularly differences in performance), and how we ended up working around those issues. The system in question leverages about 18 TB of satellite imagery, OSM, DEM and other sources and extracts information such as 2m landcover classification maps, terrain descriptors, objects such as orchards, zebra crosswalks, water towers and multiview DSMs.
During the past year we undertook an effort to transition a geospatial data processing system developed under a government research project from a small private cluster running Windows to an AWS-based environment running Linux. This talk will cover a variety of subjects including the steps taken to migrate the system, hurdles encountered with Linux and AWS (particularly differences in performance), and how we ended up working around those issues. The system in question leverages about 18 TB of satellite imagery, OSM, DEM and other sources and extracts information such as 2m landcover classification maps, terrain descriptors, objects such as orchards, zebra crosswalks, water towers and multiview DSMs.