June 2020 Summaries

Data scientists often face inefficiencies due to the significant time spent on data collection and preparation, which can occupy up to half of their workday, drastically reducing the time available for analytics and model building. Presto, a high-performance SQL query engine, addresses this issue by allowing data scientists to remain in familiar environments like R Studio or Jupyter while accessing disparate data sources without the need for extensive ETL processes. Starburst, an enterprise-ready distribution of Presto, enhances this capability by enabling fast, comprehensive data queries across various data repositories, thereby improving the speed and quality of analytics and freeing data scientists to focus on more creative and impactful projects. This approach is particularly beneficial in industries such as banking, financial services, and pharmaceuticals, where timely and detailed data analysis is crucial, and it aligns with a growing preference for using open-source tools to maximize efficiency and innovation.

Starburst has raised $42 million in Series B funding, following a successful $22 million Series A round just six months prior, bolstering its position as a cash-flow positive business rooted in open-source innovation. The funding will facilitate Starburst's mission to provide fast and affordable data access to enterprises, particularly during challenging economic times, by enhancing their analytics capabilities with minimal total cost of ownership. With new investment from Coatue, whose team includes experienced data industry leaders, Starburst aims to expand its influence and maintain its commitment to capital-efficient growth. The company, originally built on the open-source Presto platform, emphasizes its dedication to authenticity and long-term success, seeking to become a significant player in the data analytics space without excessive capital expenditure, and is actively hiring to support its growth trajectory.

Trino, an in-memory query engine, relies heavily on the underlying storage layer to optimize performance, with latency and throughput being key factors. Latency refers to the time taken to deliver data to the user, while throughput is the volume of data processed over time. To enhance Trino's performance, users should consider different storage layers like AWS S3, Azure ADLS, and HDFS for high throughput, and memory or NoSQL/document systems for low latency. Various use cases demonstrate how customers have optimized Trino queries: by moving small datasets to PostgreSQL for real-time applications, copying Oracle tables to Hadoop for frequent queries, and using memory or distributed storage for temporary tables in ETL processes. Implementing Starburst Enterprise with Trino requires a holistic approach to meet SLAs and improve user experience by leveraging appropriate storage solutions.