v4.0

This article describes the release notes for VeloDB Core v4.0.

Note VeloDB Core v4.0 was developed based on Apache Doris v2.1 (opens in a new tab) and v3.0.

v4.0.2-beta (August 25, 2024)

Improvements

Merged all features, improvements and bug fixes from Apache Doris v3.0.1 (opens in a new tab).

Bug Fixes

Fixed the issue where replaying a catalog may cause the warehouse to fail to resume.

New Features

Improvements

Merged all features, improvements and bug fixes from Apache Doris v3.0.0 rc11 (opens in a new tab) (including all the features of Apache Doris v2.1 (opens in a new tab)).
Merged all features, improvements and bug fixes from Apache Doris v3.0.0 rc09 (opens in a new tab).

Bug Fixes

Fixed the issue that after upgrading from SelectDB Core v3.0.9, background rebooting warehouse is required to restore cluster permissions.
Fixed the issue that reverse index query may get stuck.
Fixed the issue where importing into Azure Blob Storage using the s3 protocol would import too many files.

New Features

Supported multi-table materialized views (MTMV), transparent rewrite acceleration, automatic refresh, materialized views from external tables to internal tables, and direct query of materialized views. Based on this capability, materialized views can also be used for data warehouse hierarchical modeling, job scheduling, and data processing.
Supported new Variant and IP data types, and improves a series of analytical functions, making it easier to store and analyze complex semi-structured data.
Supported the ability to auto-increment columns, automatic partitioning, and centralized submission after accumulating batch requests on the server, which improves the efficiency of real-time writing of large-scale data.
Supported high-speed reading interfaces based on Arrow Flight, which increases data transmission efficiency by 100 times, making it easy to face data science and other forms of large-scale data reading scenarios.

Improvements

Merged all features, improvements and bug fixes from Apache Doris v2.1.0 (opens in a new tab).
Greatly improved the performance of blind out-of-the-box queries, achieving good performance without tuning, including further improvements in complex SQL query performance, achieving more than 100% performance improvement on the TPC-DS 1TB test data set, and query performance is at the forefront of the industry.
Greatly improved the performance of data lake analysis, with a performance improvement of 4 to 6 times compared to Trino and Spark, respectively, and the introduction of multi-SQL dialect compatibility, making it easier for users to seamlessly switch from the original system to the VeloDB Cloud data warehouse.