# Using columnar format when caching
<a name="using-columnar-format-when-caching"></a>

Spark SQL has the ability to cache tables in-memory in a columnar format. `spark.catalog.cacheTable("tableName")` or `dataFrame.cache()` function calls can be used to cache tables in an in-memory columnar format. The Spark SQL engine then scans only the required columns and automatically tunes the compression to reduce memory and CPU usage. You can use `spark.catalog.uncacheTable("tableName")` or `dataFrame.unpersist()` to remove the table from memory.