1. The verification of table data in traditional databases is schema on write (write-time mode), while Hive does not check whether the data conforms to the schema when loading, Hive follows schema on read (read-time mode), only when reading Hive checks and parses specific data fields and schemas. The advantage of read-time mode is that loading data is very fast, because it does not need to read the data for parsing, just copy or move the file. The advantage of the write-time mode is improved query performance because the column can be indexed and compressed after pre-parse, but it also takes more load time. Even for the internal table, the data format is not parsed when the data is loaded. If the data and the schema do not match, the only way to know that there is a mismatched row is when a null appears during the query.
2.hive has complex data structures (arrays, maps, structures).
3. Hive does not support real-time data processing, and the support for indexes is weak.
4.hive does not support row-level insertion.
5. High latency and large amount of data are mostly stored on hdfs.
6. Execute as mapreduce.
7.hive does not support row-level operations nor transactions.