6B.1 Scaling for Unknown Demand through the Use of Cloud Technology for the Eclipse of 2017

Tuesday, 9 January 2018: 1:30 PM
Room 10AB (ACC) (Austin, Texas)
Jebb Stewart, NOAA/ESRL, Boulder and CIRA/Colorado State Univ., Fort Collins, CO; and S. McNeil, S. Johnston, and D. Hagerty

The total solar eclipse of 2017 was one of the most highly anticipated events of the year with millions across the United States expected to migrate to the path of totality. In preparation for the event, a new eclipse enhancement was made to the experimental High-Resolution Rapid Refresh (HRRRx) atmospheric prediction system run hourly in real-time at the NOAA Earth System Research Laboratory (ESRL). The algorithm included the effects of the eclipse on incoming solar radiation and its impact on the forecast.

The combination of the eclipse and forecasts of its effects had the potential to bring significant web traffic to ESRL from a large and diverse audience. We anticipated the science community would be interested in how the weather forecast model would handle the eclipse, and the general public would want to see forecasts of where clouds could potentially impact eclipse viewing. With the existing systems already close to capacity and the threat of a surge of new users, additional resources were sought to help maintain a consistent level of service in the days leading up to and during the event.

In a very short time period, the team at ESRL and the NOAA Web Operation Center (WOC) provisioned CloudFront services from Amazon Web Services (AWS) as a way to increase capacity during this time. Through this process, the team gained valuable experience in using this service for surge capacity during high interest events. This presentation will discuss the experience of setting up the necessary services, the results of the effort, lesson’s learned and other take aways going forward with these technologies.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner