Q&A: Li Kang Explores StarRocks Cloud and Its Real-time Analytics EngineJuly 14, 2022
CloudCow recently spoke with Li Kang, VP of Strategy at StarRocks, a performance leader in real-time enterprise analytics, to learn more about the company and its solution.
CloudCow: Tell us how StarRocks came out of the work being done on the Apache Doris OLAP database?
Li Kang: The founding engineers, some of whom had worked on the Doris project before, took the initial concept of Doris and rewrote most of the modules while also adding many new features. Today, less than 20% of StarRocks code can be found in Doris.
CloudCow: How has the company evolved since first launching in 2020?
Kang: The company spent the first 6-9 months in stealth mode developing the key innovations behind StarRocks industry leading real-time analytics engine. We embarked on solving critical customer pain points to help them achieve what they could not get in existing solutions. We delivered true (seconds) real time analytics with the ability to scale to support concurrently thousands of their users.
The key innovations we developed include: our cost based optimizer, a vectorized query engine geared specifically for massively parallel needs, and the ability to support thousands of users concurrently in seconds. Since then, we built a large base of enterprise customers (over 500), including Airbnb. We are now in the process of scaling our business, adding new customers and new innovations that will continue to put StarRocks 2-3 years ahead of the competition.
CloudCow: What are some of the key features in StarRocks Cloud?
Kang: Users can get started quickly and easily with StarRocks Cloud through a user-friendly interface. With the separation of storage and compute, they can leverage various storage and compute services in the cloud for optimized performance, scalability, and cost. Streamlined setup and management means data teams can focus on providing the best possible query experience to end users and leaving the system admin work to StarRocks.
CloudCow: Is StarRocks Cloud a commercial or open source product?
Kang: StarRocks Cloud is a commercial product with the goal of enabling customers looking to quickly and easily take advantage of the industry leading real-time analytics engine. This allows them to accelerate cost effectively their time to insight at a click of a button. StarRocks source code is available on GitHub and free for on-premise deployment.
CloudCow: How does StarRocks make enterprise-scale analytics faster?
Kang: Our innovative cost based optimizer supports multi – table joins on the fly. It is designed to solve for good the slowness of distributed joins, helping deliver industry leading query execution. Our native vectorized query engine delivers an effective pipeline execution leveraging massive parallelism of CPU cores. Our intelligent materialized view provides users with transparent query acceleration, and we can support instant updates and deletes helping customers get true real-time analytics at the speed of their business.