September 2019 Summaries

SSIMWAVE is a company that works with high standards in the world of video delivery systems and analytics. They need a comprehensive database to store large amounts of data and respond quickly to ad-hoc queries. After evaluating three technologies, Apache Druid, MariaDB AX, and SingleStore, SSIMWAVE chose SingleStore due to its scalability, performance, ease of setup, direct SQL query capabilities, rowstore and columnstore support, data streaming architecture support, and wide range of integrations. The company was impressed with the out-of-the-box experience and ease of use provided by SingleStore, which enabled them to deploy their service at scale quickly and smoothly. SSIMWAVE later moved to Singlestore Helios, a high-performance, elastic cloud database service that offers improved performance and cloud services with less operations effort.

SingleStore extends its operational data platform with an on-demand, elastic cloud service and new features to support Tier 1 workloads. The company has released two new products: Singlestore Helios, a fully managed cloud database-as-a-service, and SingleStoreDB Self-Managed 7.0 Beta 2, which includes breakthrough new data management capabilities called Universal Storage. With Singlestore Helios, customers can access the world's fastest and most scalable data platform for operational analytics, machine learning, and AI on demand and at elastic scale in public cloud environments. The new product features effortless deployment and management, multi-cloud flexibility, superior total cost of ownership (TCO), and is available in limited public preview with a trial option. SingleStoreDB Self-Managed 7.0 Beta 2 introduces Universal Storage, which eliminates data duplication, reduces complexity, and cuts TCO by storing all data in a single table type. The new product also enables "system of record" capability for Tier 1 workloads through faster sync replication and incremental backup features.

Medaxion has implemented a solution for analytics problems using Looker and SingleStore. The company's CTO, John Toups, states that by combining the ease of presentation and abstraction of Looker with the technical prowess of SingleStore as the database behind Looker, they have achieved simply the best analytics in healthcare. Medaxion has improved the working lives of anesthesiologists by providing them with real-time operational information and enabling predictive analytics. The company's move to SingleStore has resulted in significant performance improvements, including a reduction in time taken for data analysis from 30 minutes to under a minute. With SingleStore, Medaxion can now easily cache large datasets in memory, reducing disk thrashing and improving overall system efficiency. The solution also frees Medaxion from operational overhead while empowering its anesthesiologist users. Medaxion has continued to innovate with SingleStore Helios, an elastic cloud service that provides the advantages of SingleStore without increasing costs. Looking forward, Medaxion is building a huge base of data to be mined and exploring potential for far-reaching change in its data architecture.

SingleStoreDB Self-Managed 7.0 introduces fast synchronous replication and incremental backups to support system of record capability, a key requirement for transactional databases. Synchronous replication ensures that transactions are persisted to disk before being acknowledged, providing an additional layer of data safety. Incremental backups allow for more frequent backups without significant performance impact. The new features enable SingleStoreDB Self-Managed 7.0 to offer a viable alternative for Tier 1 workloads and provide unprecedented design and operational simplicity, lower costs, and higher performance for a wide range of workloads.

SingleStore Universal Storage is a new vision for how databases can work, blurring the difference between rowstore and columnstore tables. In SingleStoreDB Self-Managed 7.0, rowstore tables get null compression, lowering TCO in many cases by 50%, while columnstore tables get seekable columnstores, supporting fast seeks and updates. This reduces hard choices developers have faced up to now between rowstore and columnstore tables, cutting costs and improving performance. The new system of record improvements also enable a single table type to be used for OLTP, HTAP, and analytics on arbitrarily large data sets, with optimal performance and TCO. Sparse rowstore compression improves TCO for OLTP-style applications by storing more data in the same amount of RAM, while subsegment access, hash indexes, and improved concurrency enhance columnstore tables' performance for OLTP workloads. These enhancements simplify database design and operations, enabling customers to tackle larger data management problems economically. The Universal Storage features aim to further improve performance, scalability, and simplicity, with plans for future releases including hybrid row/columnstore tables, unique hash indexes, multi-column hash indexes, ordered secondary indexes, rowstore caching of columnstore seeks, automatic adaptation of the size of the updatable rowstore segment, and a columnstore buffer pool managed directly by SingleStore.

In this webinar, Manish Pandey discusses how real-time data computing capabilities can help financial services institutions meet increasing demands from consumers. He highlights the transformative effect of machine learning and artificial intelligence in helping banks provide seamless experiences to their customers. With over 50% of consumers leaving their bank due to dissatisfaction, financial institutions must balance friction with customer expectations for a seamless experience. Pandey emphasizes the importance of leveraging digital-native companies' strategies to understand consumer behavior, as seen in the success of Google, Uber, and Facebook. He also notes that consumers are increasingly comfortable with online transactions and are expecting hassle-free interactions across multiple channels. To address these challenges, financial institutions must adopt real-time data computing capabilities, such as SingleStore, to gain a competitive edge. By leveraging analytics, machine learning, and AI, banks can predict, optimize, and forecast, while also streamlining their operations and improving customer satisfaction. Ultimately, the pace of change is accelerating, and consumers expect speed, ease, and convenience in their interactions with financial institutions.

SingleStore Studio has come a long way since its launch over a year ago with the goal of creating a visual tool to manage and monitor SingleStore clusters. The product started as a simple front-end architecture but was improved upon through customer feedback and community engagement. Since then, numerous features have been added, including Visual Explain, real-time resource monitoring, logical monitoring, and performance improvements. The latest release, 1.8.1, includes minor updates such as result tabs in the SQL Editor's output pane and fixes for a bug with tables containing date columns. Looking ahead, future updates will focus on a smoother onboarding experience, performance improvements for large clusters, and several other features yet to be announced, all of which are made possible by customer feedback and community contributions. With its growth and development, SingleStore Studio has established itself as a valuable tool for managing and monitoring SingleStore clusters.

The SingleStore Forums are seeing increased community contributions, including answers from users and valuable content, with two new Community Stars announced for the summer: Brandon Vincent and Mani Gandhi. Participation in online communities is optional, but must offer a lot to people to take off. The forums are crucial for a key SingleStore initiative, offering free use of SingleStore for small deployments, and serve as a source of help for customers and users. Community Stars are recognized for their contributions, including Brandon Vincent's documentation on Columnstore Key Guidelines and Mani Gandhi's insightful comments on Hacker News. The forums also feature discussions on technical topics, such as potential errors in pipelines and memory usage issues, alongside more lighthearted posts. As the school year resumes, activity on the SingleStore Forums is expected to increase.

In this presentation, recorded shortly after SingleStore introduced SingleStore Pipelines, two SingleStore engineers describe SingleStore's underlying architecture and how it matches up perfectly to Kafka, including in the areas of scalability and exactly-once updates. They explain that SingleStore is a scalable SQL database with a distributed system that scales out on commodity hardware, making it fast and fun to use. The engineers discuss the technical details behind SingleStore Pipelines, which are their solution to real-time streaming analytics workloads, leveraging unique properties of their existing system to create a robust and scalable workflow. They demonstrate how to create pipelines using SQL commands and show how Kafka streams can be automatically distributed across the entire cluster, enabling efficient processing and transformation of data in parallel. The engineers conclude by highlighting the benefits of SingleStore Pipelines and providing a demonstration that viewers can try themselves using a free version of SingleStoreDB Self-Managed.