As I write this article, the final hours before Labor Day are shuffling out the door along with hordes of summer vacationers. My phone isn't ringing as much. That makes this a good time to reflect on everything I've written in this column over the last several weeks.
If IBM's long-term IT strategy succeeds, what kind of technology world will we be living in five to ten years from now? Here, in three paragraphs, is my answer. We will live in a world where most of the planet's commercial computing capacity is located in giant computing utilities containing thousands of clustered and grid-connected servers. While many of the servers in these utilities will be cheap, dedicated appliances that manage a single task, the largest servers will be dynamically partitionable, capable of supporting numerous operating environments and workload types, and capable of managing themselves and entire utilities with little human intervention.
The applications running on these servers will be object-oriented and highly modular. Rather than being monolithic, they will work in concert with each other across multiple servers and networks to jointly deliver complex solutions. They will use open Web services standards based on XML to invoke and share data across terabit-speed Internet links, and grid computing technologies will ensure the performance and reliability of every transaction.
The vast majority of these applications will run on future releases of one of the two middleware stacks that are being architected for Web services and grid computing. These are the Java 2 Enterprise Edition (J2EE) stack--along with Web services and grid technologies being added by IBM and other vendors--and Microsoft's .NET stack. If IBM gets its way, most applications will run on its version of the J2EE stack, preferably on Linux and its middleware products. Of course, IBM will accommodate .NET applications as well, as evidenced by its work with Microsoft on Web services standards.
If IBM's technology world becomes a reality, what role would mid-market customers play in it? In most cases, they would be passive consumers of the IT capacity and services that the utilities provide. In some cases, those utilities would be owned by giant enterprises that are channel masters in their industry. A consortium of auto manufacturers could, for instance, create a utility for all of their parts manufacturers. In other cases, the utility might be owned by an IT services provider such as IBM or EDS. Some of these utilities would share services and data with each other by connecting their grids. Others would compete, just as wireless service providers are fighting today for the same customers. The primary task of most mid-market companies will be to pick a provider or, if they pick multiple providers, to ensure that they work amicably together.
Of course, some mid-market firms would insist on keeping all of their IT systems "off the grid" and remaining self-sufficient. Others would decide to buy or build their own grid-enabled solutions, then set up shop as small computing utilities for their business partners or their industry. However, it is likely that most small and mid-market companies would gradually outsource their IT systems to the giant utilities until the last aging server goes out the door, leaving a high-speed link to the utility where it once stood. That, plus a router connected to thin-client displays and printers, would be all that's left to support...and the utility would probably do that as well.
Will IBM's vision become a reality? There's a good chance that much of it will, since IBM's competitors are espousing similar visions. The hitch, of course, is that each vendor is putting its own spin on the vision. Among these vendors, HP-Compaq sees a world dominated by utilities (it has its own Planetary Computing project and Utility Data Center), but running on a less IBM-centric mix of technologies, including Microsoft products and HP's own systems management tools. By contrast, Sun wants to create a world in which most computing utilities are its customers--telecommunications giants and "net generation" firms--running J2EE applications on its Grid Engine software.
Then, there is Microsoft...and it is here where a very different vision for commercial computing will likely emerge. Microsoft has publicly endorsed grid computing standards and stated that it will integrate them with its XML-based Web services technologies within .NET. However, Microsoft has been strangely quiet about computing utilities. If it says anything at all, the software giant talks about computing grids enabling peer-to-peer and "participatory computing" across socialistic networks in which every device contributes according to its available resources. Instead of a world of giant utilities, Microsoft seems to be advocating a decentralized world of "IT homesteaders" who generate their own IT capacity, create their own IT services (based on .NET, of course), and connect to each other via computing grids to share both capacity and services.
Why would Microsoft propose such a vision versus the more centralized models of other vendors? Because in a world of big computing utilities, Microsoft would be the systems vendor that stands to lose the most control. That's because big utilities would be run by big, IBM-friendly enterprises that have huge investments in mainframe and Unix applications and that wouldn't think of grid-enabling them using Microsoft middleware and development tools. In IBM's world, Microsoft's efforts to work its way up from the network infrastructure into the enterprise would hit a dead end at the walls of the utilities.
In short, the IT industry will soon face a watershed event that will make the J2EE versus .NET and Linux/Windows debates look like the warm-up acts that they really are. Vendors and customers alike will have to answer the questions "Should consumers of commercial IT services hand over the production of those services to a handful of service providers? If so, who should those service providers be and what technologies should they use?" I can assure you that the battles over these questions will be fierce, for the answers will determine which IT vendors are still in business a decade from now.