Apache Hudi

What Is Apache Hudi?

Apache Hudi (Hadoop upserts, deletes, and incrementals) is an open-source data management framework designed for big data workloads. Hudi is built on top of Apache Hadoop, and it provides a mechanism to manage the data in a Hadoop Distributed File System (HDFS) or any other cloud storage system.

Hudi manages large and continuously changing datasets, such as those used in streaming applications or data lakes. It provides a way to handle incremental updates, deletes, and upserts (updates and inserts) more efficiently and in a more scalable way than traditional batch processing. Hudi uses a combination of techniques, such as columnar storage, record-level updates, and indexing to achieve this.

Apache Hudi’s approach is to group all transactions into different types of actions that occur along a timeline. Hudi uses a directory-based approach with files that are time-stamped and log files that track changes to the records in that data file. Hudi allows you the option to enable a metadata table for query optimization (the metadata table is now on by default starting in version 0.11.0). This table will track a list of files that can be used for query planning instead of file operations, avoiding a potential bottleneck for large datasets.

How Does Apache Hudi Work?

Apache Hudi enables data lakehouse architectures by providing upserts and incremental data ingestion with low-latency reads. It uses a combination of two powerful algorithms:

  • Copy-on-Write (CoW) algorithm: Allows for real-time updates and queries to data; the new records are written to a new file while the old records remain unchanged.
  • Merge-on-Read (MoR) algorithm: Merges the new records with the old ones during query time, giving the appearance of a single table or view.

Uses Cases for Apache Hudi

Data lakes -  Hudi is well-suited for use cases involving data lakes, where large volumes of data are collected and stored for future processing and analysis. Hudi can be used to manage the data in a data lake, ensuring that it remains consistent and up-to-date. With Hudi, data can be ingested from multiple sources and can be processed incrementally, reducing the need for expensive full-table scans. Hudi also provides support for column-level indexing and efficient storage formats, making it a great choice for data lakes that need to support ad-hoc querying and analysis.

Continuous analytics - Hudi can be used to provide real-time analytics on continuously changing data. It can be used to process and analyze data as it is being generated, providing insights into trends and patterns in the data. With its support for ACID transactions, Hudi can ensure that the data remains consistent and accurate, even as it is being updated in real-time. This makes Hudi a great choice for use cases involving real-time data processing and analytics, such as fraud detection, supply chain optimization, and network monitoring.

Machine learning - Hudi can be used as a data storage and management layer for machine learning workflows. It can be used to store large volumes of data, as well as manage the data as it is updated over time. With its support for incremental updates and deletes, Hudi can make it easier to manage the data used for machine learning models, reducing the need for expensive full-table scans. Additionally, Hudi can be used to store data in a format that is optimized for machine learning workflows, such as Parquet or ORC.

Apache Hudi vs. Delta Lake vs. Apache Iceberg

Apache Iceberg, Apache Hudi, and Delta Lake all take a similar approach on leveraging metadata to handle the heavy lifting. Metadata structures are used to define tables, schemas, and partitioning as well as what files make up a table.

While starting from a similar premise, each format has many differences, which may make one table format more compelling than another when it comes to enabling analytics on a data lake.

Apache Hudi vs. Delta Lake

Delta Lake is an open-source data management system that provides ACID transactions and versioning on top of Apache Spark. Like Hudi, Delta Lake is designed for managing large, continuously changing datasets. Delta Lake is integrated with Apache Spark, which means that it can be used to process and analyze data in real time. Delta Lakes also provide support for data versioning and data quality checks, making them a great choice for use cases involving data science and machine learning workflows.

Apache Hudi vs. Apache Iceberg

Apache Iceberg provides ACID transactions and versioning for large-scale datasets. Like Hudi, Apache Iceberg is built on top of Apache Hadoop, and it can be used to manage data in an HDFS or a cloud storage system. Apache Iceberg provides support for schema evolution, which means that changes to the data schema can be made without having to rewrite the entire dataset. Apache Iceberg also supports efficient indexing and query optimization, making it a good choice for use cases involving ad-hoc querying and analysis. Apache Iceberg is currently the only table format with partition evolution support.

Ready to Get Started?

Perform ad hoc analysis, set up BI reporting, eliminate BI extracts, deliver organization-wide self-service analytics, and more with our free lakehouse. Run Dremio anywhere with both software and cloud offerings.

Free Lakehouse

Here are some resources to get started

get started

Get Started Free

No time limit - totally free - just the way you like it.

Sign Up Now
demo on demand

See Dremio in Action

Not ready to get started today? See the platform in action.

Watch Demo
talk expert

Talk to an Expert

Not sure where to start? Get your questions answered fast.

Contact Us