Posted by: Jack Vaughan
Business Intellegence, Data integration
People who cover technology long enough tend to see new things in the light of past new things. If you look at MapReduce, a parallel data architecture devised by Google, it does seem like something of a threat to the unbridled growth of the RDBMS.
But someone used to this will see similarities to would-be RDBMS killers of yore, notably the object-oriented data base, which garnered a lot of venture capital, but which never required Oracle’s Larry Ellison to think about down-sizing his yacht.
As part of their due diligence on Cloud Computing, today’s software architects will take a look at MapReduce, Hadoop and other data-related schemes. Some of them may have gone through the object-oriented data base experience. Others will have worked with persistent object architectures.
What they may find is that MapReduce bears watching; that is because some things have changed in the enterprise. Notably, the RDBMS is not quite the Big Kahuna it was once. As enterprise systems grapple with mixes of numerous data types – mixes in which relational data is not the alpha and omega – MapReduce will bear a look-see. Rich Seeley reported on MapReduce, and spoke with no less venerable a technologist than Curt Monash. Monash has had his hand on the pulse of Oracle since its DB was ‘immature’ technology. By no means does he say the RDBMS is imperiled by upstart MapReduce. But his interest alone makes it worth watching, and Seeley’s story finds a other individuals articulate on MapReduce as well. Two firms, Aster Data and GreenPlum, have almost simultaneously sought to commercialize MapReduce methods.