- RDBMS Tables are Schema on Write. For each and every write operation there will be validations such as Data Types. Scale, Precision, Check Constraints, Null Constraints. Unique Constraints performed.
- RDBMS Tables are fine tuned for best of the for transactions (POS. Bank Transfers
etc) whereas Hive is meant for heavy weight batch data processing.
- where we can create indexes which are populated live in RDBMS. In Hive, indexes are typically static and when we ever add the data. indexes have to be rebuilt.
- Even though one can specify constraints in Hive tables. They are only informational. The constraints might not be enforced.
- We don’t perform ACID Transactions in Hive Tables.
- There are no Transaction based statements such as COMMIT. ROLLBACK etc. in Hive.
- Metastore and Data are decoupled in Hive. Metastore is available in RDBMS and actual business data is typically stored in HDFS.