We have looked at different formats supported by HDFS in Chapter 3, Deep Dive into the Hadoop Distributed File System. We covered many formats including SequenceFile, Map File, and the Hadoop Archive format. We will look at more formats now. The reason why they are covered in this section is because these formats are not used by Apache Hadoop or HDFS directly; they are used by the ecosystem components. Before we get into the format, we must understand the difference between row-based and columnar-based databases because ORC and Parquet formats are columnar data storage formats. The difference is in the way the data gets stored in the storage device. A row-based database stores data in row format, whereas a columnar database stores it column by column. The following screenshot shows how the storage patterns differ between these types:
