MapReduce is a model that works over Hadoop to access big data efficiently stored in HDFS (Hadoop Distributed File System). It is the core component of Hadoop, which divides the big data into small chunks and process them parallelly.
Features of MapReduce:
- It can store and distribute huge data across various servers.
- Allows users to store data in a map and reduce form to get processed.
- It protects the system to get any unauthorized access.
- It supports the parallel processing model.
Hive is an initiative started by Facebook to provide a traditional Data Warehouse interface for MapReduce programming. For writing queries for MapReduce in SQL fashion, the Hive compiler converts them in the background to be executed in the Hadoop cluster. It helps the programmers to use their SQL knowledge rather than focusing on developing a new language.
Features of Hive:
- Provide SQL type language which is called HQL.
- Helps in querying large data sets stored in HDFS(Hadoop Distributed File System).
- It is an open-source tool.
- It supports flexible project views and makes data visualization easy.
MapReduce vs Hive
|1.||It is a Data Processing Language.||It is a SQL-like Query Language.|
|2.||It converts the job into map-reduce functions.||It converts the SQL queries to HQL(Hive-QL)|
|3.||It provides low level of abstraction.||It provides a high level of abstraction.|
|4.||It is difficult for the user to perform join operations.||It makes it easy for the user to perform SQL-like operations on HDFS.|
|5.||The user has to write 10 times more lines of code to perform a similar task than Pig.||The user has to write a few lines of code than MapReduce.|
|6.||It has several jobs therefore execution time is more.||The code execution time is more but development effort is less.|
|7.||It is supported by versions of the Hadoop.||It is also supported with recent versions of Hadoop.|