SAP chief blogger Jeff Nolan comments on Martin Lamonica's story on "Software Stack Wars"--the notion that enterprise software vendors are obsessed with building their own pre-certified stacks, from operating system and middleware to applications and databases, or some variation. I wrote about Oracle moving in that direction, based on CEO Larry Ellison's remarks about adding Linux to have a more complete Oracle controlled stack. The logic is simple--customers mimimize cost and complexity in areas where differentiation is mimimal among parts of the stack and vendors maximize profit from services contracts.
One reader expressed the choices as follows:
The question ultimately will be is the customer better served by tightly linked and highly cohesive software stacks (I'm guessing not) or by lightly linked and highly cohesive products that are best in class.
I'm not sure about the best in class or best of breed as a way to reduce complexity and integration woes, but highly cohesive, yes. Jeff points to a Talkback that takes a different cut at the superstack concept.
The article talks about the large enterprise IT vendors' focus on "stacks" as if it's a recent rediscovery: but the truth is that "best of breed" strategies haven't been popular since the mid 1990s. That's practically ancient history in the world of IT. SAP, IBM, Microsoft, Oracle, Sun, BEA et al have been playing the stack game since way back. All that changes is the labels on the boxes.
What's interesting to me, is for how long the tired and entrenched stack plays of these vendors are going to hold up in a world where enterprise customers increasingly look to the power of the network as the driver of value and innovation.
Yessir, in tomorrow's world value will come from leverage of software service networks, not software product stacks. [Ital mine]
Nobody is really expecting to get into a position that locks out any other other companies, the reality is much more that various components interoperate in our client’s datacenter.
We are well past the days when stacks were truly monolithic in nature and static in their linkages, in fact you could reasonably suggest that the forward thinkers in this business stopped thinking that way as early as the CORBA days and more broadly following the advent of true web applications.
True to a large degree, but Microsoft and Oracle, at least, seem to have more of an Alexander the Great mentality--annex whatever part of the software stack you can to increase the surface area--and others dream about superstacks, especially in an era where the profits are in support, maintenance and service delivery.
If you are SAP, and your are an applications company as well as development platform for third parties, like Salesforce.com and AppExchange, you want outside developers to extend platform reach. Jeff mentions how Shai Agassi, head of products and technology at SAP, said that his company is "not a platform company, we’re an application company." I would characterize SAP as an applications platform company, given how NetWeaver is positioned as the foundation of an application and integration ecosystem built around SAP's proprietary technologies.
The fact that NetWeaver (or what other enteprise software vendors will provide in their "stacks") is service-oriented, not welded shut or monolithic, speaks to the idea of leveraging software service networks. Software will be more "composable," but at the same time, those SOA-based vendors will not resist the urge to add to their stacks to varying degrees to gain more share of the customer IT spend...