Vectorization in Hive is a performance optimization technique that allows Hive to process large amounts of data more efficiently. It works by processing multiple rows of data in a single iteration, instead of processing each row individually. This can result in significant performance improvements, especially for analytical queries that operate on large datasets.

Vectorization is enabled by default in Hive and can be controlled using the hive.vectorized.execution.enabled configuration property. When vectorization is enabled, Hive uses a vectorized query execution engine to process data.

Here is a simple example of how vectorization can be used in Hive:

-- Create a table
CREATE TABLE sales (
  item_id INT,
  sale_date STRING,
  sale_amount FLOAT
);

-- Load data into the table
LOAD DATA INPATH '/data/sales' INTO TABLE sales;

-- Use vectorization to calculate the sum of sales for each item
SELECT item_id, SUM(sale_amount)
FROM sales
GROUP BY item_id;