When new technology trends appear, it's hard to separate the fads from the long-term solutions. Well, hard might be the wrong word: Impossible might be better.
The truth is, you can guess but you won't know until the early adopters report back on their experiences. Even so, what works for one company may not work for another, as Rick Sherman points out in his DM Review column, Data Integration Advisor, this month.
In his August column, Sherman looks at five major - some might say majorly hyped - trends in data integration. Sherman is a data warehouse and business intelligence consultant with 20 years of data work history, so his educated guess is backed by real-world experience.
Here's my Cliff Notes version of Sherman's predictions: SOA: Sherman describes SOA as "very, very hyped," but "also very real and will become pervasive." That said, he predicts current SOA efforts will fall short of their expectations. We won't see the real benefits until the second or third generation of SOA apps. He compares SOA to client/server and Web, saying both were described as killer apps but were actually enabling technologies. He thinks SOA is headed down the same road.
As a side note, Nick Malick addressed a similar question this week in the last of his three-part series on Selling SOA (and why you shouldn't). Malick questions whether SOA is a disruptive technology and concludes it is not. Along the way, he also gives us an inside peek at how Microsoft's IT division does its job and works with the business side.
SaaS: It's working for applications, but Sherman doesn't think it will work for the data integration market, due to security and performance issues. He believes it could work for the business intelligence market, though points out performance will still be the key challenge to its success.
Open source software (OSS): A long way off from being a real-world trend, particularly in companies with existing BI and data warehousing solutions. Still, open source BI and data warehouses may find ground in small firms that haven't invested heavily in these technologies.
Master data management (MDM) and customer data integration (CDI): Overhyped. MDM/CDI is less about software than people and defining processes. If you're trying to buy a solution, you'll be disappointed.
For a different opinion on MDM solutions, check out my recent Q&A with IBM's senior product manager, Karen Leightell.
Sherman goes into much more detail, so if you have the time, do the homework and read the full column, which is quite short, but well-stated.
While you're on DM Review, check out the article, "BI, Process and Integration Trends," which looks at business intelligence information management and enterprise information management - think legacy information systems. It includes a discussion of two problems these approaches share in common: data integration and process.