Shell—not ‘big’ data, Odata!

Shell CIO Johan Krebbers unveils new enterprise IT architecture at OSIsoft conference including in memory computing with SAP Hana and multiple endpoint support from ‘Odata’ open data protocol.

In a wide ranging presentation to the OSIsoft user conference in San Francisco last month, CIO Johan Krebbers outlined Shell’s plans for an overarching IT infrastructure spanning the whole company, ‘from drilling to retail.’ Shell’s new infrastructure is built on three pillars—the cloud, an enterprise data model and endpoint flexibility. The cloud is now its preferred way of buying IT with a preference for software as a service—acquiring an ‘end to end’ solution. While in the recent past Shell was a strong supporter of Microsoft Windows and browser-based endpoints, ‘those days are going rapidly.’ Today’s users work with iPhones, tablets or, in the near future ‘smart glass’ devices. Today, people ‘live and die with their smart phones,’ now a critical part of the environment that brings its own security and authentication issues.

For Krebbers, ‘big’ data is irrelevant. Data, of any size, needs to be transformed with analytics to be actionable. Today, production, drilling and retail are all components of an enterprise-wide real time environment. In every corner of the enterprise there is a move from look-back reporting to forward looking analytics that predict future performance.

Data is getting bigger though, as sensor counts increase and as the ‘internet of things’ envisages more connected devices. A modern seismic survey can generate 10PB of data which may expand tenfold during processing—a true big data challenge. Shell’s global, remote operations span unmanned platforms that mandate a ‘data driven’ environment. Systems and users need more trustworthy data. Data quality is improved as data is made visible. The other quality driver is the enterprise data model—an initiative to move towards a single version of truth and systematic naming of devices and business objects.

Tools of Shell’s big data trade include an SAP/Hana in-memory appliance with a petabyte of main memory. Less mission-critical data may be stored in SQL databases or in Hadoop if current use cases are unclear. Data in Landmark and PI is combined with historical data—a key enabler of non conventional ‘factory’ drilling.

The enterprise data model ‘Edam’ is under development by a team of 20 data modelers along with rules for data ownership and governance. A data services layer provides access to multiple data sources, hiding complexity. Endpoint access leverages the Odata protocol. Shell’s SmartApps (OITJ April 2012) are being retooled around Odata. Ultimately all PI data sources will be represented in a 3D virtual reality ‘living’ model of Shell’s assets that will last throughout the plant’s lifetime. Watch the OSIsoft video of Krebbers’ presentation.

Click here to comment on this article

Click here to view this article in context on a desktop

© Oil IT Journal - all rights reserved.