Comments on: HadoopDB: An Open Source Parallel Database for Analytical Workloads

By: Hyunsik Choi

Hyunsik Choi — Fri, 31 Jul 2009 18:15:19 +0000

I also think that a hybrid system will play an alternative role between anti-RDBMS and parallel RDBMS. Actually, proponents of anti-RDBMS claim that RDBMS already reached a limit for satisfying manifold demands that incur in various environments. I agree some of their advocates. However, MapReduce is not Swiss Army Knife, so we would need the alternative to take ability of both.

By: woorung

woorung — Fri, 31 Jul 2009 02:40:20 +0000

I think that the final winner between anti-RDBMS and parallel RDBMS will be a hybrid system which aims at integrating MapReduce into RDBMS.Actually, GreenPlum and HadoopDB are doing that.Both of them are from RDBMS advocates since they have lots of knowledge and experience through RDBMS research.Especially, the hybrid system needs SQL-like query analysis & optimization to manipulate distributed DBMSs with MapReduce.In this point, I think that RDBMS advocates cannot help defeating anti-RDBMS advocates, unfortunately.Nevertheless, they won't lead IT industry & market due to somewhat high cost.To reduce the cost, most of people want to take advantage of open sources.I think that the debates were ended up with HadoopDB.Woohyun Kima creator of open source coord which provides C++ MapReduce framework and distributed key-value store.

By: Hyunsik Choi

Hyunsik Choi — Thu, 30 Jul 2009 17:18:38 +0000

#chunglab HadoopDB Releases! http://bit.ly/ZntJ5