MapReduce (NoSQL): A major step backwards

databasecolumn.vertica.com

4 pointspj2 comments

I see this stupid article pop up every couple months. There are a few things wrong with it:

1) Map/Reduce is an algorithm, not a datastore.

2) Map/Reduce is geared towards problems that SQL does not work well for.

3) They are not talking about Map/Reduce for most of the article. They are talking about schema-less document based data stores, like BigTable or CouchDB.

4) This is FUD published by a bunch of RDBMS consultants. They have a vested interest in shooting down competing technologies. This is pretty obvious from their _complete_ misunderstanding of the technology that they are attacking.

5) I dont understand why the poster specifically tagged (NoSQL) in the header since this predates the NoSQL "movement", and has nothing to do with the experimental RDBMS.

tl;dr : move along folks, nothing worthwhile here.

The post says that they are talking about the nonrelational database systems in general, that's why i added noSQL to the header.

I posted it because it is a post that supports the contrary position to many of the pro-nosql comments going around lately. Just to add both sides of the equation... to maybe, help get to truth.

Reply on news.ycombinator.com