Cloud-Based Mission Platforms: Availability Analysis of the GeoCloud System

As Cloud computing becomes integral to satellite data processing, ensuring the high availability of these platforms is essential. This poster presents an in-depth availability analysis of GeoCloud, a cloud-based mission platform developed by the GOES-R Ground Segment Project and hosted on Amazon Web Services (AWS). GeoCloud provides important Mission Services to the GOES-R Mission Partners, including the National Weather Service and instrument teams, facilitating weather prediction and supporting post-launch testing and anomaly resolution activities.

GeoCloud hosts a range of services, including GeoProduct Access Service (GPAS), GOES-R Level Zero Archive Service in the Cloud (GL0ASc), Electronic GOES-R Export Service in the Cloud (eGRESc), GeoStream, and the ABIPLTc tool. These services operate within the same IT Security Boundary as the on-premises GEO Ground System and are supported on an 8x5 basis. Although they are non-real-time with no latency requirements, their availability is crucial for mission success. The availability requirement for GeoCloud is set at 98.9%, allowing for a maximum of 96 hours of aggregate downtime annually.

Unlike traditional on-premises systems, the availability analysis of cloud-based systems like GeoCloud presents unique challenges. Since the underlying hardware is managed by the cloud provider, we rely on their Service Level Agreements (SLAs) and performance metrics. This lack of direct control and visibility over the hardware introduces a layer of uncertainty that must be accounted for in our analysis. Building an accurate availability model and performing the necessary calculations is challenging, as is accounting for all the details and various components of the system.

Our analysis confirms that GeoCloud meets this stringent requirement, ensuring reliable access to mission-critical data and tools. This poster will detail the methods used in the availability analysis, the model that was built, the results obtained, and their implications and lessons learned for future cloud-based mission platforms.