Tuesday, June 26, 2007

Thoughts on GE's next generation system based on Oracle - part 1

A number of people have asked about my thoughts on GE's announcement back in March about a next generation system based on top of Oracle technology (note that in this post, GE stands for General Electric, not Google Earth!). This was covered by Joe Francica at All Points Blog under the title "Cutting out the GIS Middleman", by Susan Smith at GISCafe in an interview with Robert Laudati, and in an article by GE. Here are my thoughts ... (warning - this is a rather long post, in fact it was getting so long I have decided to split it into two parts, and the first part is still very long!).

I'll give a quick bit of history on Smallworld for all you "neo" people and those who haven't had any involvement with GIS in the utility industry. Smallworld is where I worked from 1992 to 2002, and during that time we grew from being a small startup in the UK to the global market leader in GIS for utilities and communications (according to Daratech), with revenues of around $100m. Smallworld was bought by General Electric in 2000. Smallworld introduced some radical new ideas in the early 1990s, many of which have now become common practice across the industry - I will talk more about some of these in the future, but Charlie Savage gives a good summary of his perspective here. In the 1990s, Smallworld had a clear technical lead in the utility industry, but in the early 2000s both ESRI and Intergraph introduced new systems (ArcInfo 8.0, now ArcGIS, and G/Technology), and the playing field is now much more even, with no clear leader, in my opinion. The Smallworld product remains very robust and scalable and has very rich functionality, but increasingly GE has been suffering in new sales situations because of the fact that Smallworld is based on its own proprietary language (Magik) and database (VMDS, Version Managed Datastore), while its two primary competitors have more modern and mainstream software architectures. Actually both Magik and VMDS still have some great technical strengths, especially the latter, but as geospatial technology has moved more into the mainstream, it has become increasingly hard to convince the market of the benefits of buying a "proprietary" solution, and in my opinion this has been the primary driver for GE to develop this new product set on top of Oracle.

The most obvious interesting thing about this announcement, which others have commented on also, is that it really reinforces the notion that geospatial technology is becoming absorbed into mainstream IT - many people have talked about this for a while, myself included. The way it was explained to me by someone from GE is that when they first sat down with Oracle to discuss collaboration on this project, they thought that there would be three layers of software: Oracle technology at the back end, a new "GIS" layer in the middle, and specific utility applications built on top of that. But when they looked at what Oracle now offers, including a map viewer, network model, version management (workspace management in Oracle terminology), etc, they concluded that there were really just two layers: Oracle, and the specific utility applications. I don't think this point is especially interesting to customers, actually: whether you regard the solution as 2 tiers or 3 tiers is a bit of an arbitrary distinction - in either case you need software from two vendors, Oracle and GE, and whether specific bits of functionality come from one or the other is not really significant. If GE reduces the cost of their software because they don't have to develop as much functionality themselves, then customers will be interested, but I haven't heard any discussion about that!

The reason this is interesting is from a general industry perspective, because it suggests that it will be harder to have a successful business which focuses purely on the middle "GIS" layer, without delivering applications on top of that which solve specific business problems. There are now many free or cheap options for drawing a map on a screen; you no longer need a specialized and expensive piece of software to do that. Companies like GE and Intergraph sell both geospatial platform software and vertical industry applications, and both are now putting more emphasis on the vertical applications where (in the right areas), they can still show high business value and therefore justify relatively high prices, and less emphasis on the basic geospatial capabilities, many of which are becoming commoditized. ESRI has of course focused heavily on delivering a horizontal GIS product, and has a large partner ecosystem which provides vertical applications on top of this. They have a very dominant position in this horizontal GIS space, and the fact that many of their traditional competitors are focusing more on vertical applications may mean that they can increase their hold there - but nevertheless they are seeing significant pressure on various parts of this space from Oracle, Google, Microsoft, Yahoo and open source solutions. So it will be interesting to see whether ESRI keeps its horizontal focus or also starts to move into more vertical solutions over time. There would be some challenges in this strategy, in particular the issue of potentially competing with its partners - though this is a common problem for platform software companies and is in many ways a natural evolution that many companies go through. We went through this at Smallworld, starting as a platform company with partners and then moving more into vertical applications, and Oracle is experiencing this also, especially with several of its recent acquisitions (more on that in a future post).

OK, so does this new architecture give GE a competitive advantage? Not really, in my opinion. The first major argument in favor of it is that all your data is stored in a standard relational database, which helps with integration, administration, security, etc (see my 1990 article which outlines these benefits - these concepts are not new!). The second major argument is that you can use standard development environments (Java-based, in this case). So GE addresses the concerns that the market today has about its existing "proprietary" solution - but both ESRI and Intergraph have provided solutions based on mainstream databases and development environments for a number of years, so these things will not be differentiators for GE - they are playing catch up in this regard. Oracle likes the fact that the solution is based purely on the Oracle stack - and so customers who have committed to the whole Oracle stack will also see that as an advantage. But on the other hand, customers focused on a Microsoft architecture either on the client or middle tier will see this approach as a disadvantage - and my general feeling is that more utilities probably fall into this category. Those who are not too religious about their IT strategy (probably the majority) will focus on the functionality provided by the main three vendors more than the system architecture.

So, in summary on part 1: I am pleased to see GE make this announcement, as I had personally pretty much written off the possibility of them investing in a "next generation" system. I am happy for the friends I still have at GE that there is investment in the future, and I think it will be good for the industry if GE can make this new product into a strong competitor to ESRI and Intergraph in the utility market - and if they can bring forward the strengths of the current Smallworld system then it will be. The announcement is interesting because it shows that you can now develop complex geospatial applications directly on top of Oracle without needing a traditional "GIS". While I think this is of somewhat academic interest to most customers, it is more significant in terms of what vendors in the geospatial industry will look like in future. I personally don't think that the market will see the new system architecture as a competitive advantage, except in situations where organizations have very strong Oracle religion - I think it will be seen as more of a catch up exercise by most people, which negates the perceived weakness of Smallworld's current "proprietary" architecture.

In part 2 I will talk about the real reason why this product could be a significant jump forward for the utility industry, which really hasn't been highlighted in the GE announcements or in any of the commentary I've seen. And I'll talk about why this same factor could be the reason that the product fails. And last but not least, I'll talk about some of the challenges which GE faces in positioning the new product with regard to the existing Smallworld products.

1 comment:

rkgeorge said...

Hi Peter,
Great post. The old GIS vendors seem to be in a retraction mode. Instead of looking at a next generation they appear to be retreating into a middle tier, business logic, stance where both the DB and the Client side can eventually be pluggable commodities.