Hadoop Vs OLTP

Hadoop Vs OLTP

Can Hadoop be used as an OLTP database ?

No. Hadoop is not a replacement for a transaction RDBMS.

Hadoop is a framework for distributed processing to deal with large volumes of data. Hadoop  does not provide any random access to data stored in its files, so if you are really interested in saving and accessing your data, check HBase. Having said that, HBase again is not a replacement for a traditional transactional RDBMS and is generally used for sparse big data sets. Moreover, for banking, Hadoop is not the use case.

hBase has been the de-factor standard for OLTP in Hadoop (assuming you use the same nodes for HDFS that you use for hBase, which you should).  Cassandra also fits into this category.

What is OLTP ?

OLTP (On-line Transaction Processing) is characterized by a large number of short on-line transactions (INSERT, UPDATE, DELETE). The main emphasis for OLTP systems is put on very fast query processing, maintaining data integrity in multi-access environments and an effectiveness measured by number of transactions per second. In OLTP database there is detailed and current data, and schema used to store transactional databases is the entity model (usually 3NF).

Leave a Reply