Migrating U.S. Geological Survey National Seismic Hazard Models to the Cloud
Session: Forthcoming Updates of the USGS NSHMs: Hawaii, Conterminous U.S. and Alaska [Poster]
Type: Poster
Date: 4/28/2020
Time: 08:00 AM
Room: Ballroom
Description:
The USGS delivers National Seismic Hazard Models (NSHMs) both as published reports and static data. The latter, along with dynamic hazard calculations, are delivered via web-services and applications, the use of which has been increasing over time. Under the current architecture, all supported NSHMs are delivered via a single application, the Unified Hazard Tool (UHT), with a limited number of backing web-services that are hosted on USGS infrastructure. However, each model release requires more resources to be allocated, which is unsustainable as models become more complex and require more memory. To address this, the USGS has started to utilize cloud-based solutions, specifically Amazon Web Services (AWS), which allow for immense flexibility given a multitude of computing resource configurations available to end users. To facilitate cloud-based deployments, we now wrap individual models and associated web-services in Docker containers that can be deployed to Docker swarms or AWS using the AWS Elastic Kubernetes Service. This deployment model provides improved scalability support because additional containers can easily be spun up to accommodate any increased web-application computational load. A container-based deployment model also improves running hazard calculations and web-services locally because end users no longer have to deal with complex issues related to dependencies and compilation. At this time, we are also refactoring web-applications to use the Angular framework. Angular simplifies the building, testing, bundling and running of USGS web applications. This framework is being adopted by other USGS development groups, for example earthquake real-time products and geomagnetism, which will improve long-term support of our applications and will provide greater opportunities for internal and external collaboration.
Presenting Author: Brandon S. Clayton
Authors
Peter M Powers pmpowers@usgs.gov U.S. Geological Survey, Denver, Colorado, United States Corresponding Author
|
Brandon S Clayton bclayton@usgs.gov U.S. Geological Survey, Denver, Colorado, United States Presenting Author
|
Migrating U.S. Geological Survey National Seismic Hazard Models to the Cloud
Category
Forthcoming Updates of the USGS NSHMs: Hawaii, Conterminous U.S. and Alaska