May 2019 Summaries

SingleStore has helped Kurtosys, a digital experience platform for the financial services industry, to speed up analytics and gain faster performance. Kurtosys initially used Couchbase but found it challenging to work with due to its lack of SQL support and difficulty in running queries against structured data. After researching and selecting SingleStore, a fully distributed database that supports MySQL wire protocol, Kurtosys was able to use the change data capture (CDC) process built into SQL Server to keep SingleStore's copy of the data up to date. This allowed Kurtosys to dedicate SQL Server to transaction support while running queries against SingleStore, resulting in improved performance and scalability for analytics and transactions. The new platform created by Kurtosys uses SingleStore as the database of record, supporting JSON data and providing outstanding performance with flexible storage options. Kurtosys is now looking forward to using the new features of SingleStore to power the growth of its platform, offering cutting-edge platforms that will continue to push SingleStore forward.

SingleStore has achieved a significant milestone by successfully running three distinct database benchmarks - TPC-C, TPC-H, and TPC-DS - on the same scalable database, demonstrating its ability to handle both transactional and analytical workloads simultaneously with excellent performance. This breakthrough showcases SingleStore's unique capabilities as a relational database that can be fully distributed and horizontally scaled, making it an attractive option for operational analytics where live data is streamed into a data store to drive real-time decision-making. The company has optimized its query optimizer for both transactional and analytical workloads, achieving impressive results in benchmarking tests, including high sync replication performance, near-linear scaling, serializability, and high availability. SingleStore's capabilities are being utilized in production by top-tier companies across various industries, including finance, gaming, and energy.

With the advent of cloud computing, organizations face new opportunities and challenges in managing their resources. To take advantage of public and private cloud resources without being locked into a specific cloud or infrastructure, container solutions like Docker can be used to develop, deploy, and manage software. Recently, Kubernetes has added support for stateful services, and SingleStore has developed a beta Kubernetes Operator that allows users to easily manage containers across public clouds and existing infrastructure. The operator is currently experimental and in beta, but it offers benefits such as easy deployment and installation, elastic scaling, and smooth upgrades, making it an attractive option for organizations looking to leverage the power of Kubernetes with their database needs.

SingleStore has spent six years developing a fast and capable database to meet the needs of modern businesses, which require data infrastructure that can handle an increasingly broad set of requirements and scale well at scale. The company's latest release, SingleStoreDB Self-Managed 6.8, offers improved analytics performance and advanced security for Hadoop environments, while the upcoming beta release of SingleStoreDB Self-Managed 7.0 previews new transactional capabilities for system of record applications. SingleStore has demonstrated its ability to meet or beat industry benchmarks in a range of workloads, including those that require both fast ingest through transaction processing and low-latency queries for reports and dashboards. The company's database is designed to support operational analytics, which involves a specific combination of key features such as fast ingest, fast processing, and low-latency queries. SingleStoreDB Self-Managed 6.8 includes improved query performance, including a 5x speedup for certain types of queries, as well as advanced security capabilities for Hadoop environments, including Kerberos support and wire encryption.

SingleStore has demonstrated strong performance on the TPC-C, TPC-H, and TPC-DS benchmarks, showcasing its ability to handle both transactional OLTP workloads and data warehousing tasks. The company's operational analytics capabilities allow it to run analytical queries across large datasets with consistently high performance. SingleStore's scale-out architecture and use of single instruction, multiple data (SIMD) instructions enable it to achieve dramatic benefits in performance compared to traditional databases. By running on industry-standard hardware without special-purpose optimizations, SingleStore has proven its value proposition for operational analytics and cloud-native applications. The company offers a free version with support for all enterprise features, making it an attractive option for businesses looking to adopt a scalable database solution.

In the latest version of SingleStoreDB Self-Managed, version 6.8, several query processing capabilities have been enhanced to improve performance and user experience. The marquee feature is just-in-time (JIT) compilation, which speeds up query runtimes on the first run of a query, now turned on by default. This feature can greatly improve the user's experience running ad hoc queries or when using applications that cause a lot of new SQL statements to be run. Additionally, performance improvements have been made in certain right and left outer joins, Rollup and Cube, as well as sub-select without an alias and extended Oracle compatibility for date and time handling functions. New array functions for splitting strings and converting JSON data are also added. Furthermore, the resource governor has been extended to better manage resources for more operations, and new built-in functions have been introduced, including SPLIT() and JSON_TO_ARRAY(). Overall, these enhancements aim to provide improved performance, user experience, and compatibility with various databases and applications.

Many companies have invested heavily in Hadoop infrastructure, but it falls short for certain use cases. To address this, SingleStoreDB Self-Managed 6.8 introduces Kerberos and wire encryption for HDFS pipelines, providing comprehensive security through full standards-based authentication and over-the-wire data delivery.

Hadoop was introduced in 2008 as a solution to organizations' big data problems and gained popularity due to its promise of scalability and cost-effectiveness. However, many industry experts argue that Hadoop may not be the panacea for big data problems it was hyped up to be. The main reasons include skills shortage, high costs, complexity in getting data out, hiring software engineers is expensive, and Hadoop is not optimized for real-time decision making. These limitations have led to a decrease in Hadoop adoption and growth within existing enterprises.

Kubernetes has revolutionized application development, deployment, and maintenance, and the newly released SingleStore Kubernetes Operator simplifies managing an enterprise-grade operational database. The operator is certified by Red Hat to run on OpenShift, offering flexibility across hybrid, multi-cloud, or on-premises environments. With the operator, developers can deploy and manage a fully distributed database that deploys and scales instantly, configured quickly using the native SingleStore API. This innovation allows customers to deploy SingleStore on any public or private infrastructure more easily, simplifying management tasks for DevOps and administration teams.

Intel’s new Optane DC persistent memory offers a cost-effective and high-capacity alternative to traditional RAM, providing a 25% average performance reduction compared to using only traditional RAM, while being much less expensive than traditional RAM. The new technology can be used in memory mode, where the total Optane capacity acts as addressable RAM capacity, allowing servers to support larger rowstore tables with significant cost savings and easier management. This is especially beneficial for SingleStore customers running large rowstore tables, enabling them to upgrade their effective RAM capacity much further and make projects with 10TB or more of data feasible. The technology also has potential applications in other workloads that use rowstore tables in SingleStore, but its benefits may vary depending on specific use cases.

The customer, who previously ran their business on two databases, AWS RDS and Druid.io, saw significant cost savings after moving to a single database, SingleStore. The new monthly cost is about $31,000, resulting in a savings of $62,000 per month or 66%. This move also brought increased performance, greater concurrency, and easier database management. SingleStore's flexibility allows it to run in various environments, including public clouds, private clouds, on-premises, and mixed environments, reducing the risk of "cloud lock-in". The customer was able to easily migrate from AWS RDS Aurora to SingleStore due to its compatibility with MySQL wire protocol. Additionally, SingleStore can augment existing databases like Aurora, allowing for faster performance and improved concurrency. Overall, this move resulted in a dramatic cost savings and strategic benefits for the customer.