PPDM, POSC Geoshare - how many standards do we need? (April 1998)

PDM's editor Neil McNaughton muses on recentdevelopments in the data modeling arena and asks - is it important to the buyer?

This month's announcement from PI(Dwights) endorsing the PPDM data model over POSC's Epicentre is the latest in a protracted series of alternating skirmishes and rapprochements between the two organizations over the last few years. To recap for those of you who may have come late to this edifying spectacle, the story starts around 1991 when POSC, located in Houston, (but with strong backing from European National oils) and the Canadian Public Petroleum Data Model set out independently to build the "mother of all data models". POSC got some forward thinking purist data modelers on board, whereas PPDM was founded by commercial practitioners of data modeling, including Finder Graphics Corp. - later to be subsumed into the GeoQuest organization.

Middleware

PPDM focused on producing a working database, defined at the level of tables and rows, with real-world compromises both in terms of scope and de-normalization. At the same time POSC's approach can be categorized as an attempt to model the real-world of E&P in as complete and uncompromising a manner as possible. Another important difference in concept was the way in which applications would use these databases to share the data. In the case of PPDM, all applications were to access the data through direct SQL access, whereas POSC developed a sophisticated scheme using a "middleware" layer, the Data Access and Exchange layer - which was to allow better separation of application and data, better future-proofing and all sorts of other benefits.

The first observation on both of these efforts is that neither has met with the initial expectations. Despite vendor claims, there are no commercial products that use either PPDM or POSC in a way that could be said to correspond to the initial intent. Both models are used in different products - even in the same product - yet no "compliant" application can access data in another vendor's equally "compliant" database. There have been two major attempts over the years to bring POSC and PPDM together - to no avail - while the two major vendors appear to use the standards issue as a marketing football, with both GeoQuest and Landmark settling on POSC's Epicentre data model, but with neither offering interoperability with the other via a DAE.

The other runner in this race is Geoshare, which proffers yet another data model. Geoshare has not had the benefit of as much promotion as POSC or even PPDM, but it is the de-facto standard for the exchange of data between E&P applications, and therefore is as near as we have got today with interoperability. Next month's PDM will report from POSC and PPDM member meetings as well as the Geoshare user group, we'll be giving a "state of the nation" analysis of data modeling and models, and who is using them and for what. Meanwhile we invite you to read the contributed article by Mark Chidwick (Panther) and Bruce Rodney (Exprodat) for another view on where we are on the long and winding road to interoperability.

In the interim, PI's abandonment of POSC contrasts with the apparently enthusiastic claims for POSC compliance made by both Landmark and GeoQuest. It is hard to extract a straightforward editorial line from these developments. On the one hand, data gathering and use may mandate different models. On the other hand, how many standards can we manage? As a variant on my conclusion last month, "buy not build" is all very well, but if we end up with complex and hence costly systems, who is going to pay for their maintenance? The oils of course, in more ways than one.