November 30, 2022
Hadoop File Formats and its Types

Hadoop File Formats and its Types

It seems like wherever we turn, new sources of information are being added to the mountain of information we already have. This lesson will cover each of these sections in depth, since the information is broken down into several sections.

We’ll examine the various Hadoop file formats in the next section.

Types of Hadoop File Formats

There are four distinct Hadoop file formats that may be used to construct tables in HDFS applications like Hive and Impala.

Text files make up the Sequence File.

Information stored in files using the Avro encoding

Parquet is a file format used for storing and transferring

Examine each file type used by Hadoop in detail.

Text files

Text files are the most basic and straightforward of all file formats. Since much of the code is separated by commas or tabs, it may be read and written in any programming language.

If you’re using a text file format, you’ll need to save numerical values as strings, which takes up more room on your hard drive. It’s also challenging to represent binary data, like a picture.

Sequence File

The sequencefile format can be used to keep a binary copy of a picture. When compared to a textual format, the binary container format is superior for storing key-value pairs. Contrarily, sequence files cannot be read by humans.

Avro Data Files

The Avro file format is useful for its space-saving binary encoding. It has strong backing both within and beyond the Hadoop community.

The Avro file format is ideal for archiving critical data for extended periods of time. Several programming languages are supported, including Java, Scala, and others. Schema information can be included in the file to guarantee its readability at all times. The evolutionary process is flexible enough to accommodate a new schema. Avro is the preferred format for most Hadoop users since it may be used for a variety of different purposes.

Parquet File Format

Cloudera and Twitter, with with others, created the Parquet columnar format. Spark, MapReduce, Hive, Pig, Impala, and Crunch are just some of the many frameworks that offer support. Similar to how Avro metadata is incorporated in the file, the schema metadata is as well.

In his work, Dremel explains how Google’s Parquet file format takes use of sophisticated improvements. As a result of these enhancements, we can now accomplish more with the same amount of storage capacity. This Parquet file format is often regarded as the most effective for adding numerous records simultaneously. Observing patterns is the foundation of several optimization techniques. What follows is an explanation of data serialization and how it functions.

What is Data Serialization?

Data serialization is the practice of representing information as a string of bytes for the purposes of computer storage. You may use it to either store information locally or transfer it across a network.

How would you recommend serializing the number 123456789, for instance? When saved as a Java String, a Java int takes up just 4 bytes, but a Java String takes up 9 bytes.

Data serialization framework Avro is widely used in the Hadoop ecosystem and beyond. There is also support for RPC, or Remote Procedure Calls, which ensures compatibility with the coding environment without sacrificing speed. Take a look at Table 1 to learn more about the data types that may be stored in and retrieved from Apache Avro and Avro Schemas.

Supported Data Types in Avro

The following is a list of the elementary data formats that Avro can accommodate.

As a synonym for “null,” this term describes a situation in which no value has been provided.

A boolean represents a binary value.

Commonly used in computer programming, the Int:int data type is a 32-bit signed integer.

A 64-bit signed integer, long is the longest kind of integer.

The symbol for a single-precision floating-point number is “float.”

The symbol double denotes a floating-point number with two decimal places of accuracy.

Information is often stored as a string of 8-bit unsigned bytes, or bytes.

string Unicode text is stored in what is called a “string.”

See how Avro schemas handle the many complicated data formats out there.

A record is a type that has been specified by the user and has one or more fields with predefined names.

Enumeration; a list of possible values that has been predetermined.

array: An array contains zero or more elements of the same kind.

A map is a collection of related information, organized by key and value. Strings are used as keys, and values can be any number of different things.

Union: A value that unites a group of types is called a union.

This phrase designates a certain quantity of 8-bit unsigned bytes within a file.

Take into account how Hive’s use of the Avro schema facilitates the development of a database table.

Hive Table and Avro Schema

The following is an illustration of a table in the Avro schema, which may be compared to a table in the Hive schema.

Specifically, the following: , (id INT, name STRING, title STRING);

The following example illustrates how to assign a default value to a schema in Avro.





“name”:”id,” “type”:”int,” and other similar terms.

the variables name” and type” are both set to a string,

The title is the name of the type, and the string is the type of the data.


Operational Activities of Avro





“name”:”id,” “type”:”int,” and other similar terms.

In the example above, the name would be “name,” the type would be “string,” and the default would be “opennel.”

The title, the type, and the default value are all strings.] [name=”title,” type=”string,” default value=”bigdata.”]


Here we will look at how Sqoop is utilized by the Avro and Parquet file formats.

Avro With Scoop

To move information between Hadoop and an RDBMS, the Avro and Parquet file formats necessitate the usage of Sqoop, a program specifically built for this purpose.

For example, you may use Sqoop to transfer data from a relational database management system to HDFS in Avro format and vice versa. For this task to be completed, the —as-avrodatafile argument must be included to the Sqoop command.

Hive can read and write any Avro format. However, Impala does not support nested or sophisticated Avro types like enum, array, fixed, map, union, or record (nested).

Here we’ll discuss utilizing the Sqoop tool to import data stored in the Parquet file format.

Parquet With Sqoop

With Sqoop, you may load data in Parquet format from a relational database management system into Hadoop File System (HDFS) and export data in Parquet format from HDFS to a relational database management system (RDBM). To do this action, you must add the -as-parquetfile option to your Sqoop command line. Here we’ll go through how to bring MySQL data stored in Parquet File Format into HDFS.

Importing MySQL to HDFS in Parquet File Format

MySql data in the Parquet file format may be imported to HDFS in two ways: by making a new table in Hive, or by reading the files directly. Let’s have a more in-depth examination of them.

Creating a New Table in Hive With Parquet File Format

You can make new Parquet tables in both Impala and Hive. With Impala’s LIKE PARQUET function, you may import column information from an existing Parquet data file. The picture below shows this in action. A new table and a new order are being made so that data stored in the Parquet file format may be read. Due of their binary nature, parquet files can be challenging to understand.

Reading Parquet Files

Downloading the parquet-tools package (described further below) will allow you to read parquet files.


This course provides an introduction to the Hadoop file types.

  • The HDFS file system allows users to import text files to generate tables in Hive and Impala.
  • Examples of data storage formats include sequence files, Avro data files, and Parquet file formats.
  • Data serialization refers to the process of representing data in a computer’s memory as a sequence of bytes.
  • There are several tools and modules in the Hadoop ecosystem that make use of Avro, a data serialization mechanism.
  • Sqoop allows for the import of data in the Avro and Parquet file formats into HDFS.
  • Data may be exported to a relational database management system from file formats including Sqoop, Avro, and Parquet.


Of course, it’s crucial to learn about the many Hadoop file formats available. Hadoop file formats are only one of several subtleties associated with Big Data and Hadoop.