← Back to Sessions

Seismic Networks in the Cloud

Description: 

Traditional earthquake data processing is carried out on dedicated hardware in a data center run by the operator of the seismic network. Such specialized data centers have to be carefully planned, and hardware resources must be purchased in advance or constantly aggregated/modified/updated. Traditionally, private leased lines, satellite, and/or cellular networks build the communication backbone. Over the years, the internet has become a significant carrier for seismic data directly from the field as well as between institutions. With the evolution of the internet as the backbone of international data exchange, multi-tenant data centers started to evolve in many scientific fields, leading to virtualized cloud computing.
However, implementing a process-based seismic real-time data center with 24/7 operation requires additional considerations. Nevertheless, the modular infrastructure of a virtual data center allows the flexible scaling of processes when more data has to be processed. And cloud data centers allow easy horizontal scaling without limits for computational and storage resources. Adopting cloud services is neither straight nor easy, and scaling and flexibility come with a price. The radical change towards a pay-per-use infrastructure versus the traditional investment makes obsolete several consolidated paradigms. The budget shift imposed by cloud services from capital investment to operational costs requires adapting the traditional budgeting processes. In a conventional data center (covered by the initial investment), practices with negligible expenses could be costly in the cloud environment where customers are charged per use. The presentation compares systematically and financially two cloud configurations: one for a small scientific network with a few data users and a very large national network with many stakeholders, from the network operators to emergency officers.

Session: Geophysical Data Analysis in Cloud Computing Environments [Poster]

Type: Poster

Date: 4/18/2023

Presentation Time: 08:00 AM (local time)

Presenting Author: Mathias Franke

Student Presenter: No

Invited Presentation: 

Authors