I have often been perplexed, over the last ~25 years, by assorted waves of technologies and/or market positioning meant to suggest that the extended relational database management system market was somehow past its prime. I covered this general topic in a Burton Group Data Management Strategies blog post a few months ago, before I parted ways with Burton Group; skim that post for more details on the broader DBMS market reality check dimensions.
While many people still believe that DBMSs are only useful for basic text and number data/application domains, and that open source DBMSs such as MySQL are "good enough" for a wide range of database management needs, in reality the leading commercial DBMSs -- which for Linux and Windows servers means mostly Oracle Database and Microsoft SQL Server, these days (with IBM DB2 in 3rd place) -- are expanding to address "the other 80%" of information that hasn't traditionally been managed in DBMSs, ranging from basic file management to elaborate XML documents and advanced, composite data types (e.g., for spatial application domains).
And while some people assert that extended relational DBMSs are needlessly complex "bloatware", I think you'll find the people making that type of assertion are in many cases trying to sell you some type of self-proclaimed "next generation" specialized database system, and not being entirely objective about the DBMS incumbents.
It's critically important to understand that the advances embodied in Oracle's Exadata and Database Machine announcements today apply to *all* information management domains, not just high-end data warehousing, and that overall information management market momentum is shifting more toward extended relational DBMSs than into any sort of specialized alternative model.
Especially with today's privacy and security realities, and with advances such as no-compromises DBMS-based XML content management and XQuery, in many ways the database management market is just getting started, and the implications, in terms of future business opportunities for DBMS-centric platform vendors such as Microsoft and Oracle, are pretty amazing...