Hive versus relational databases

hive is so similar to relational databases that there is always an illusion in hive learning that hive is a database, not a database. Hive is the client side of Hadoop, with HDFS at the bottom, and the execution engine is MapReduce, which is executed on Hadoop and, in other words, a layer of Hadoop’s client package.

1. Data update

• hive read more write less
• mysql usually needs to modify

frequently
2. Data delay

• mysql usually executes in seconds
• hive for a longer time:
  • hive query, there is no index, need to scan the whole table, so the delay is high
  • mapreduce when the hive is executed, there will be a shuffle, shuffle to drop the disk, the delay is high

3. Data size

• hive data scale is large
  • hive is stored in HDFS and built on clusters. You can add machine vertical expansion
• mysql has storage bottlenecks
  Mysql store on disk