2A.2 Leveraging HPC Technologies to Accelerate GOES-R Product Processing and Distribution

Monday, 11 January 2016: 1:45 PM
Room 355 ( New Orleans Ernest N. Morial Convention Center)
Brad Brown-Bergtold, Harris Corporation, Melbourne, FL; and J. Gundy

Handout (1014.5 kB)

The GOES-R mission presents multiple challenges for producing and distributing products at rate. GOES-R will produce raw data at a rate of 100Mbps and Level 0/1/2+ data at a rate of 16.1 TB per day. Product latencies can be as low as 1.8 seconds for solar products and 23 seconds for imagery products. Given satellite image processing is generally highly parallel, Harris utilized High Performance Computing (HPC) techniques and technologies to develop a parallel processing framework that provides the flexibility to configure workloads based on data partitioning. This framework is optimized for high throughput, low latency transactions that are highly reliable and is hosted on a Beowulf style cluster of commodity hardware. Both the framework and underlying hardware are designed to scale beyond 300 percent
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner