April 2016 Summaries

We are delighted to announce that SingleStore has raised more than $36m in a Series C round. The oversubscribed round includes new investors REV Capital, Caffeinated Capital, with full pro rata participation from earlier investors including Accel (who led our Series B), Khosla Ventures, and Data Collective. Our investors have looked at the SingleStore technology and concluded that SingleStore is the leading real-time database platform for analytics and operational workloads. Large enterprises want SQL, fast and easy to use, which SingleStore provides, enabling them to transform their business and respond quickly to trends. The company powers real-time applications such as billing systems reaching 6M updates per second and logistics analytics that previously took 24 hours now run daily in under an hour. With the growing need for real-time insight, SingleStore is well-positioned to expand its talented team, fund product development, and provide a real-time database platform to transform businesses.

The terms rowstore and columnstore have been widely adopted in database management for online transaction processing (OLTP) and online analytical processing (OLAP) workloads, respectively. However, a nuanced approach is required to determine when to use each type of storage based on the specific characteristics of the workload. Rowstores excel at random reads and writes, while columnstores are better suited for sequential reads and writes. Despite common myths, columnstores can be faster than rowstores in certain scenarios, such as workloads with sequential scans or computationally analytic but operationally constrained constraints. Hybrid Transactional/Analytical Processing (HTAP) workloads, which blend both OLTP and OLAP characteristics, often benefit from using a combination of rowstore and columnstore indexes. Ultimately, the choice between rowstores and columnstores depends on the specific requirements of the workload, including the types of operations performed, storage constraints, and performance considerations.

SQL is the dominant approach to data processing, and SingleStore has stayed true to this approach. The company is exploring in-memory architectures as a key technology for future computing. Lambda architecture is another area of focus, with SingleStore helping to break down its complexities. The industry is also seeing a resurgence of interest in traditional database technologies, such as caching and high-speed counters. Geospatial analytics is becoming increasingly important, particularly in real-time applications. SingleStore's ecosystem fits into a larger market landscape, including HANA and Hadoop, and the company has partnerships with Oracle to help customers get the most out of its technology.

At Strata+Hadoop World, a major event in data management and analytics, SingleStoreDB Self-Managed 5 was announced by CEO Eric Frenkiel, offering breakthrough performance on various workloads. The new release features an LLVM code generation architecture, query optimization improvements, columnstore performance enhancements, and SQL features such as a new `EXPLAIN` command. Additionally, Eric shared his vision for real-time processing and predictive analytics with the Strata audience, highlighting the potential of combining technologies like Kafka, Spark, and SingleStore to drive business insights. A tutorial session was also held by Kellogg's JR Cahill, showcasing how they integrated SingleStore into their data analytics workflow.