Since the initial fork from Doris three years ago, the StarRocks team has re-written about 90% of the code base to improve performance, stability, usability, etc. In the past three years, the StarRocks team has replaced the query optimizer with a brand new Cost Based Optimizer to eliminate de-normalization, implemented a Vectorized Query Engine to improve query performance, designed a Primary Key Data Model to better handle real-time analytics scenarios, released Intelligent Materialized Views to simplify data pipelines, and rolled out many other breakthroughs.
While it seems Doris has been trying to play catch up since last year, StarRocks obviously has a 2+ year head start.
The two products seem to be headed in different directions. While Doris continues to focus on real-time analytics, StarRocks has added data lake query capabilities. The latest benchmark using TPC-H shows StarRocks is 3x to 5x faster than traditional data lake query engines like Trino. This makes StarRocks the first platform for Data Warehouse, Data Lake, and Real-time analytics use cases.
The StarRocks community is led by CelerData and consists of 200+ contributors across the globe. CelerData also offers the cloud native managed service of StarRocks, called CelerData Cloud.