August 2015 Summaries

SingleStore is a relational database management system that offers both familiar SQL interface and leading-edge features such as memory-optimized data storage, distributed architecture, and software-only deployment. It stores data in memory for rowstore tables and on disk for columnstore tables, with options to store data in compressed format on flash/SSD/disk. SingleStore ensures durability by writing logs and full database snapshots to disk, and supports transaction processing with configurable parameters. The system is architected as a cluster of aggregator and leaf nodes, allowing for scalability and flexibility. Licensing is based on total RAM capacity of the cluster, rather than CPU cores, enabling customers to maximize resources without licensing overhead.

The in-memory database market is experiencing a high demand for real-time capabilities, with analytics and data capture being top use cases. The survey also highlights the growing interest in Apache Spark, with over 50% of respondents planning to evaluate it in the upcoming year. Spark Streaming is the most widely adopted Spark library, allowing companies like Pinterest to structure real-time data and measure user engagement in real-time. However, common challenges such as log processing bottlenecks, subpar performance, high costs, and complex queries for analytics are still prevalent, highlighting the need for an in-memory database that can provide solutions to these issues. SingleStore has addressed some of these challenges with features like JSON support, a productivity tool called SingleStore Loader, a Spark Connector, and converged processing capabilities through hybrid transaction and analytical processing (HTAP).

The text discusses the importance of online ALTER TABLE operations in relational databases, which allow for schema changes without downtime in production. It highlights that many traditional relational databases have poor support for this feature, but SingleStore achieves true online ALTER TABLE capabilities by minimizing impact on running workloads. The database's design ensures minimal blocking and system resource usage during the process, making it suitable for large-scale enterprise users who rely on painless schema changes.

Visit our SoMa headquarters this Wednesday, August 19th for an exclusive meetup where Drew Paroski will share his expertise on building compilers. Compilers play a crucial role in maximizing application performance by translating programming languages into machine code, resulting in significant speed improvements. At the meetup, Drew will outline key considerations for building the best possible compiler, including identifying performance goals and evaluating full-custom approaches versus alternatives. The event also features a database speed test contest with a prize for the best performance, as well as a happy hour with heavy hors d'oeuvres and a Q&A session with Drew. This is an opportunity to learn from someone who has co-created the HipHop Virtual Machine and Hack programming language, and to network with like-minded professionals in the field.

Mesosphere DCOS is a distributed operating system designed to span all machines in a datacenter, allowing for easy deployment of applications across the entire system with simple commands. SingleStore can be easily deployed on DCOS due to its distributed and memory-optimized design, providing users with flexibility to scale computation and storage capacity by adding nodes. Deploying SingleStore on DCOS is straightforward using the DCOS command line interface, requiring only a few simple steps to get started. The DCOS web interface provides an intuitive way to monitor and manage SingleStore clusters, allowing users to start, stop, and delete clusters as needed. With Mesosphere DCOS, SingleStore offers flexible deployments across various infrastructure options, including physical machines, VMs, private clouds, public clouds, and Docker containers.

Scala is a programming language that combines object-oriented and functional programming paradigms, allowing it to be used in both traditional Java-like styles with mutable state and highly functional manners. This hybrid approach makes Scala appealing for big data pipelines, where concurrency and scalability are crucial, as it reduces the need for mutable data and naturally handles changes in state. With its widespread adoption by Twitter and other large enterprises, Scala has become a popular choice for building scalable applications, particularly with Apache Spark, Kafka, and other frameworks that leverage its functional capabilities. The success of Scala is expected to contribute to a shift towards more functional approaches in the programming landscape, with tools like SingleStore enabling a hybrid object-oriented/functional approach to data management.

The SingleStore CEO Eric Frenkiel will be speaking at the Gartner Catalyst Conference on August 11 about building real-time data pipelines with Apache Kafka, memory-optimized Apache Spark, and SingleStore. The conference also features a happy hour event, complimentary Gartner reports, games, and giveaways, including a chance to win an Estes Drone by playing a reaction test game and wearing the latest SingleStore t-shirt. SingleStore will be exhibiting at booth #518, offering demos of their products MemCity, Supercar, and Pinterest, as well as meeting with potential customers ahead of time.

SingleStore has released its Community Edition on the Amazon AWS and Microsoft Azure Marketplaces, making it easier for customers to deploy and manage their SingleStore databases in the cloud. Many companies already use SingleStore in the cloud for various applications, including real-time charging platforms like VCare, which uses SingleStore to power its online platform supporting Mobile Virtual Network Operators. With the release of Community Edition on AWS and Azure, users can launch a "cluster in a box" on a single virtual machine or run a performance benchmark that supports over 1 million inserts per second. The quick start guides for launching SingleStore on these platforms are now available in the documentation, along with a Cloud Formation Template generator for multi-vm clusters on AWS. This move expands access to SingleStore's in-memory technology, which can be used for various purposes such as powering applications and monitoring datacenters.

The latest version of SingleStore Ops, version 4.0.34, is now available for download, offering new features to accelerate productivity for users. The enhanced superuser account allows for more intuitive management and access control, while the installation directory improvements make it easier to navigate to SingleStore installation directories and access their content. Additionally, this release introduces more metadata about hosts in the cluster page, including High Availability pairings and display preferences, as well as customizable upgrade files that give users control over which version of SingleStore they want to install.

SingleStore was recognized as a strong performer in the 2015 Forrester Wave evaluation for in-memory database platforms, receiving high scores in categories such as scale-out architecture, performance and scale, and product road map. The report validates SingleStore's approach to in-memory computing, citing its benefits of great performance when working with massive volumes of data. As adoption of in-memory databases grows, companies are turning to SingleStore for fast, scalable infrastructure to support predictive applications that require instantaneous access to hot data. The Forrester Wave concludes that customers building new transactional and analytical applications should look at SingleStore for its extreme performance and low-latency access capabilities.