What SQL does Meta use?

Meta uses a custom SQL-based solution called Presto (now known as Trino) for much of its large-scale data querying needs. Presto is an open-source, distributed SQL query engine developed by Meta to run fast queries on massive datasets across distributed databases.

Here’s how Meta uses SQL and SQL-based technologies:

1. Presto (Trino):

   • Presto was initially developed by Meta (Facebook) engineers to allow fast, scalable SQL queries across large data lakes and distributed storage systems like Hadoop. It enables interactive querying of large datasets without requiring data to be moved or pre-aggregated.
   • Use Cases: Meta uses Presto for analytics, data science, and querying large amounts of data across various storage systems.
   • Source: Presto Website

2. MySQL:

   • Meta also uses MySQL as part of its relational database systems for handling transactional data. MySQL is typically used for smaller, fast-access storage needs where structured data is essential.
   • Enhancements: To manage the massive scale of Facebook's data, Meta customizes MySQL to fit its specific needs. One example is using RocksDB as a storage engine to optimize MySQL for read-heavy workloads.
   • Source: Meta’s MySQL Use

3. Hive:

   • Meta also leverages Apache Hive, a data warehouse software built on top of Hadoop, which provides an SQL-like interface to query large datasets stored in distributed storage systems.
   • Use Cases: Hive is used for large-scale batch processing of data within Meta’s infrastructure.
   • Source: Hive

Summary:

• Presto (Trino): Used for querying large, distributed datasets.
• MySQL: For relational database needs with custom enhancements.
• Hive: For large-scale data processing.

These SQL-based solutions allow Meta to handle the massive scale of its data operations across its platforms.

Sources:

• Presto Documentation
• Facebook’s MySQL Customizations
• Apache Hive