Christopher Bradley (IPL) and Ken Dunn (BP) presented a paper on ‘Data Modeling as a Service (DMaaS) at BP’ at the Data Modeling Association (DAMA) annual meet in San Diego this month. BP has a huge dispersed IT system as witnessed by some statistics from its Digital and Communication Technology (DCT) division. BP has 250 data centers (these are consolidating to three ‘mega data centers’), 80,000 desktops, 6,000 servers, 7,000 applications including 33 SAP instances, 26 ‘major’ data warehouses from SAP and Kalido all running across 30 petabytes of spinning disk.
Reasons to model
BP models big time! To capture business requirements, promote data reuse and assure quality, to bridge between business and technology and to assess the ‘fit’ of packaged solutions. Modeling helps BP identify and manage redundant data and provides ‘context’ for projects—facilitating systems integration.
Vision
BP’s vision is for data and information that are managed as an easily accessible shared corporate asset. This means integrated, enterprise-wide global data, a single view of customer and product master and ‘real-time’ processing where needed. The DCT unit has a role to play in coordinating initiatives that target data quality, visibility and ownership.
Decentralized
BP’s decentralized environment has led to a situation where there is no single subject area model and few modeling standards. Historically a project focused approach led to documentation getting lost. Continuity of resources has proven problematical as much work is out-sourced. Another driver for change was the need to avoid BP’s ARIS process modelers developing data models independently.
The solution
The solution, co-developed by BP and UK-based IPL was for ‘self service’ administration—with registration required for use of Embarcadero Technologies’ ER/Studio modeling tool. When a model is approved, an automated publishing process moves models to BP’s Data Modeling Environment SharePoint repository.
Community of Interest
A modeling ‘community of interest’ (COI) shares business cases, best practices and proposes ‘domain directives’ for data modeling at BP and also avoids ‘re-inventing the wheel.’ The COI can also influence Embarcadero’ product development. The COI also handles certification of internal and suppliers.
2008 on ...
Looking forward BP is working on a model validation service, with promotion of ‘approved’ master data models and industry standard models such as PODS*. The COI is also driving a ‘quality model culture’ and seeking to develop cross domain modeling governance. BP is also anticipating a ‘services-oriented architecture (SoA) world’ where ‘definition of data and results from services are vital. Other facets of BP’s ongoing modeling effort address reverse engineering of legacy models in business warehouse information ‘cubes’ and ‘business object universes.’
* Pipeline Open Data Standard—www.pods.org.
© Oil IT Journal - all rights reserved.