Speaking at the Strata/Hadoop World big data conference in San Jose, California last month, CGG’s Marco Ippolito presented CGG’s high performance computing capability which includes a 10 petaflop GPU-based machine at one processing center (see page 3). Even so, increasing processing workload has created technical challenges, notably of processing to storage bandwidth. Another issue is the marketing ‘noise’ coming from the ‘big data’ movement.
Salvation has come from Objectivity and its ThingSpan platform for ‘information fusion’ solutions that enrich big data assets with real-time data streams. CGG uses ThingSpan to enable ‘parallel seismic data visualization and analysis.’ ThingSpan deploys a Lambda architecture and runs natively on the Hadoop file system as a Yarn application. Apache Spark is also used for workflow and data transformation. The result is a data store that spans seismic volumes, petrophysical data and reservoir models. CGG’s Jason Workbench common data model also ran.
CGG GeoSoftware SVP Kamal Al-Yahya commented, ‘Over the past ten years, Objectivity has helped CGG leverage various IT advances in our commercial software. ThingSpan now provides us with a low-risk avenue into the Hadoop ecosystem and increases deployment options for our clients.’
Read the presentation here.
This article originally appeared in Oil IT Journal 2016 Issue # 4.
For more information or to comment on this topic email here.
This recent, 181 word, article is currently for subscribers only. To request a copy, click here. This is a discretionary offer, restrictions may apply.