If any topic brings forth visions of blind people groping at the elephant, the fundamental architecture of how GIS fits into enterprise applications is it! It sure got the sparks flying last May at the Location Technology and Business Intelligence conference.
Dean Stoecker, president of SRC is among those who believe "It's wrong to invest in a spatial platform." Spatially enabled technologies should be integrated into existing platforms. Jack Pellicci, Vice President with Oracle, spoke along similar lines, claiming that business needs "integrated management of all types of data: it's not just about location."
Pellicci then proceeded to expand on this thesis, taking it into controversial territory. "Spatial is not special; location is too important to be left to specialists. The things you need to cross that chasm [between non-spatial and spatially-enabled data] are in place today.You don't have to go to a specialist.Someday, we may not need GIS tools as we see them today."
In some of my previous columns, you will recognize how opposed this vision is to the views that "wetware" really matters. Jack Dangermond (ESRI) rose immediately to the bait. "I don't have clarify on what the technological solution is, but in the presentations we saw people who are special, with cartographic tools or geographic knowledge.'Spatial people are not special' is bull****!"
It would be easy to understand this exchange as a challenge to the GIS platform developer issued by the systems integrator and the enterprise database vendor.Each is jockeying for his place in the market. But I think there's more to it than that. There is, as yet, no shared vision of how businesses will make best use of spatial technologies while there is sharp disagreement over how to adopt them. Is the value mainly in having the technology and underlying data available to everyone throughout "the enterprise," or is the value in having access to those who can use the technology and data to help make "actionable" business decisions? It probably depends on the business.
Pellicci stuck to his guns. "People are looking for fewer maps
and more answers," getting people the data they need in a timely way.
Paul Overberg, database manager at USA Today, weighed in on Dangermond's
side, with a qualification. "Where there's a lot of intellectual
capital, as in law offices, ...there may be less need of ERM and more of
GIS."
In subsequent remarks, Dangermond seemed to be seeking a middle ground. "Take accountants, for instance: their systems won't make them go away. ...GI-intensive workflows will continue to use GIS, while at the same time there will be enterprise systems."
Dangermond sketches his vision of where the technology is going. Small company solutions will move to Web services and pay by the use: "one-tenth of a cent for a map." This vision of Web services contrasts with the "forcing-it-all-into-one-can model" of a database management system, which Dangermond termed the "1990's approach." This is a new system, more than just spatially enabling the enterprise. "We are really encapsulating knowledge." Geospatial services and business services will be integrated together within a Web services environment and "3D GIS visualization will improve our ability to communicate."
Dangermond was the first and only participant to offer an explicit definition of an enterprise system. There are four basic tables in an enterprise database, he told us: money, product, personnel, and customer tables. To those add "business logic" and applications: this is the enterprise system.
Concerning those applications, George Moon, CTO of MapInfo, offered that "You need access to intelligent information at the back end, but it doesn't mean you have to get rid of the fat client [the application]." Both he and Dangermond appear to be seeking a place where their products can exist within a general business intelligence technology infrastructure.
Now, though, according to Dangermond, another table is emerging: the geography table. He asked us, "Is location another part of tabular data, or are data a part of the geographic layer? It depends on the business focus. This 'abstraction of geographic things' is driving ESRI into the 21st century."
With that somewhat enigmatic remark, we left the subject. Do we "GIS the enterprise" or "enterprise the GIS?" I'm not sure exactly what these phrases (coined by Dangermond) mean, but one of them seems to consist largely of adding "fat" GIS applications to existing business IT and the other to subsume spatial information storage, processing, and analysis capabilities within the IT infrastructure. The first will be in most need of specialists and be more amenable to "point" solutions, whereas the second eschews specialists and purports to make spatially-enabled information available to (and presumably usable by) anyone in the enterprise without needing a specialist's help.
If our eggs are in the application basket (ESRI, MapInfo), we would like to choose the former. If we are a player in the enterprise market (SRC, Oracle), we would like to choose the latter. It is unclear which is better for the customer, or whether one model must necessarily win out. I think Dangermond intends for ESRI to survive and prosper in either model. I wish him well.