In biology, we are taught that survival favors diversity. Organisms that reproduce without variation die out during periods of rapid change, while organisms that show variation in feature tend to survive and adapt. Likewise, ecosystems consisting of relatively few homogenous species thrive only when conditions stay static.
Does IT diversity create a competitive advantage in the business application ecosystem? Predictably, large vendors with vertically integrated stacks argue that mixing software components is a Bad Thing. These vendors claim that reducing the diversity in the application stack leads to better efficiency and maintainability.
For most of recent history, Microsoft has had the best integrated application development stack, and many - particularly smaller - enterprises adopt the Microsoft stack in its entirety. However, while Microsoft has a significant presence in custom and web applications, it does not offer credible enterprise ERP or CRM solutions.
Conversely, SAP offers completely integrated ERP applications, but not an application stack suitable for building custom or web applications.
It's only been recently that a single vendor - Oracle - has been able to offer a completely integrated IT stack that includes both the application platform and the applications themselves.
For most of the past decade, Oracle has strategically acquired technologies with the aim of providing a "Complete, Open and Integrated" application stack. While some might dispute that the result is truly open, it is indisputably more complete than any competing enterprise technology offering.
Acquisitions such as BEA (Java Application Server), Hyperion (Business Intelligence), Siebel (CRM), and Peoplesoft (ERP) already provided Oracle with unparalleled coverage of enterprise software. This year's finalized acquisition of Sun Microsystems gives Oracle a leading UNIX operating system (Solaris), as well as server and storage hardware offerings. Oracle also has integrated strategic open source components such as Oracle Virtual Machine (based on the Open Source Xen technology) and Oracle Enterprise Linux (based on the Red Hat Linux code line).
Oracle's president, Charles Phillips, recently argued that enterprises adopting the Oracle stack in its entirety can reduce the management overhead that a "mix and match" approach involves. Since Oracle certifies and tests the entire stack, incompatibilities between various layers in the stack are minimized. While a unique combination of best-of-breed technologies might, in theory, offer the most advanced functionality, the costs of provisioning and maintaining these customized combinations, in practice, are too high, according to Phillips.
It's true that CIOs are attracted to the simplicity of single vendor support - ‘One throat to choke' - as they say. It's also true that the single integrated stack may have lower deployment and maintenance costs.
Against these advantages, CIOs need to consider the lack of leverage that often accompanies vendor lock-in. When you standardize on a vendor, you can place yourself at that vendor's mercy when it comes to renewal and maintenance agreements. Big integrated stacks also have big, integrated price tags.
The single integrated stack also has a higher risk of obsolescence when confronted by disruptive technologies. When a disruptive technology forces rapid change to a specific level in the application architecture, massive consolidated stacks are unlikely to adapt as quickly as mixed stacks, which already assume some abstraction between the application and the database layers.
Wholly integrated stacks may offer some compelling economic advantages over the short term. However, over the longer term, large integrated stacks are less adaptable than loosely coupled best-of-breed hybrid stacks, and may become expensive and inefficient legacies when technology paradigm shifts occur. Many CIOs will choose to hedge their bets by adopting a mix of tightly integrated vertical stacks and multi-vendor best-of-breed technologies.