SQL on Hadoop

Run SQL over big data.

SQL on Hadoop reaches big data on HDFS with SQL. Hive compiles SQL into MapReduce; Presto runs an in-memory distributed query instead.

SELECT customer_id, COUNT(*)
FROM hive.orders
GROUP BY customer_id;

Hive turns SQL into MapReduce.

customer_id, COUNT(*) is selected from hive.orders. Rows are grouped by customer_id; aggregate functions are computed per group.

SQL on Hadoop — details

Hive is slow but handles very large data; Presto is fast but bound by memory. They pair up: Hive for ETL, Presto for interactive analysis.

Select 10 rows from the hive.orders table: SELECT * FROM hive.orders LIMIT 10.

Show example solution

SELECT * FROM hive.orders LIMIT 10;

Hive catalog

SELECT * FROM ____.orders LIMIT 10

Show answer

SELECT * FROM hive.orders LIMIT 10

Next lesson: SQL vs NoSQL