CODEWITHSUNDEEP
hadoopmapreducebusiness-logic
Matthew Moisen
Matthew Moisen

Reputation: 18289

Is there any Business Logic that Hadoop/Mapreduce Cannot Handle?

I currently have five tables that receive replicated data on a frequent basis. PL/SQL is used to transform, cleanse, and aggregate this data based on complicated business logic. The results are then placed in three reporting tables.

I am interested in using MongoDB instead of a relational database for this. In order to do so, all of the PL/SQL business logic must be translated into another language, such as Java.

However, it has been suggested to me that Hadoop will be a better alternative than MongoDB. I am now learning about Hadoop and Mapreduce, but I get the feeling that Hadoop is used for analyzing data (deriving statistical value from data, i.e. correlations between customer actions) not for transforming data based on business logic.

In order to replace the as-is system with Hadoop, I assume that all of the PL/SQL business logic will have to be translated into a mapreduce function. Is this assumption incorrect?

Regardless of my personal situation, is there any business logic that Hadoop (and mapreduce) cannot handle, that PL/SQL or Java can?

Upvotes: 1

Views: 512

Answers (2)

Thirupathi Nerella
Thirupathi Nerella

Reputation: 1

I can confirm that you can use MongoDB+Hadoop. All you have to do is, place entire transformation logic from all packages at a side then identify input,output and flow (dependencies) then convert to map reduce.The major task is to identify key,values,rules from transformation logic. I hope you can use pig also. Use Sqoop for extraction.

Upvotes: 0

mat_vee
mat_vee

Reputation: 85

First,MongoDB is type of NoSql database and Hadoop is kind of parallel processing framework. If you really have tera/peta bytes of data it is justifiable to use parallel processing framework. And consult your data that is it okay to go with Nosql database + parallel processing framework.

  1. You can convert PL/SQL business logic into mapreduce functions
  2. Hadoop can handle any business logic lot loss if indexing,centralized processing.(For non-transactional systems)

Also I will suggest you look into hive it might help you.

Upvotes: 1

Related Questions