More on the MapReduce debate

As previously noted, I think MapReduce is a big deal. Tonight I chatted with MapReduce's most famous non-fan, database management legend Mike Stonebraker, and he laid out his reasons for not liking MapReduce. However, it's far from clear that those objections apply to all MapReduce implementations. While they are quite plausible about Hadoop, they seem less applicable to Greenplum's and Aster Data's new database-integrated MapReduce implementations. And so, as much as I admire Mike, I'll have to disagree with him on this one, and suggest that MapReduce should be seriously considered in at least the four overlapping application domains where it already has traction:

  • Text tokenization, indexing, and search
  • Creation of other kinds of data structures (e.g., graphs)
  • Data mining and machine learning
  • Data transformation
Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.

Copyright © 2008 IDG Communications, Inc.

IT Salary Survey 2021: The results are in