July 2015 Summaries

Drew Paroski, co-creator of the HipHop Virtual Machine (HHVM) and the Hack programming language at Facebook, has joined SingleStore as a champion of in-memory computing. Prior to his tenure at Facebook, Drew worked as an engineer at Microsoft and IBM. At Facebook, he was part of a three-person team appointed to rebuild the software layer of the company's web-tier to support over one billion new users, building HHVM to improve PHP performance and co-authoring several patents in just-in-time compilation, dynamic programming languages, and runtime data structures. Drew is now focused on innovating with SingleStore's query compiler, aiming to make distributed SQL fast and improving the overall performance of applications. He sees opportunities for growth at SingleStore, particularly in its ability to provide real-time big data analytics capabilities to companies large and small.

The trend in working with data today is towards using databases as intended, storing data separately from applications, adding structure to unstructured data, making data accessible and utilizing real-time stream processing frameworks. This approach can save time and money by reducing the need for frequent updates and improving efficiency. In contrast, some common mistakes include calling a database representative for scaling support, rather than adding nodes with more flexible databases, and relying on outdated knowledge of past trends instead of staying current with present-day techniques.

The data market is rapidly growing, with predictions that it will double in size from $60 billion in 2014 to $115 billion in 2019. The Internet of Things (IoT) spending is expected to reach $1.7 trillion in 2020. In-memory technology is gaining popularity due to its ability to provide capabilities needed by companies to keep up with the growing data market. The main reasons for this adoption include declining costs in memory and infrastructure, high-performance use cases, improved return on investment (ROI), and enhanced data persistence options. These factors are driving the growth of the in-memory database management system (IMDBMS) market, making it an attractive solution for organizations looking to improve their business insights and reduce operating costs.

The SingleStore Meetup is scheduled for July 28th at 6pm, focusing on building a successful sales team through hypergrowth. Jim Herbold, former Executive Vice President of Global Sales at Box, will share insights on scaling a sales team and go-to-market strategies during hypergrowth. The meetup features a happy hour, optional database speed test contest, main presentation by Jim Herbold, and Q&A session. Attendees can learn from Jim's experiences growing the Box sales team to 400 people and taking revenue from $1 million to $174 million.

With a few simple commands, users can launch a SingleStore cluster on Mesosphere DCOS, a commercially supported product built on top of Apache Mesos that serves as a datacenter abstraction layer. SingleStore is a relational, in-memory database optimized for multi-machine deployment and leverages the Marathon scheduler to deploy the software tightly integrated with DCOS. The package needed to deploy SingleStore on DCOS can be found in the Mesosphere Universe repository, allowing all DCOS users to install SingleStore anytime simply by running `dcos package install memsql` from a DCOS console. SingleStore supports various deployment topologies, including bare metal machines, virtual machines, Docker containers, public cloud platform providers, and Mesosphere DCOS itself.

This post discusses the benefits of migrating from NoSQL databases to relational databases, specifically highlighting the features of SingleStore that make it an attractive option for companies looking to switch. The author shares their experience with migrating from MongoDB to SingleStore and provides a step-by-step guide on how to get started with SingleStore's native JSON support. Key features highlighted include JSON column types, JSON field predicates, persisted columns, updating JSON document fields "in-place", manipulating JSON arrays, finding the last element of an array, and performance considerations. The author emphasizes the importance of focusing on domain modeling and functionality before optimizing for performance.

We are excited to exhibit at the Data Science Summit on Monday, July 20, in San Francisco, where we will showcase our MemCity demo and give away a cool t-shirt as well as an Estes ProtoX Mini Drone. The event is a non-profit gathering that brings together researchers and data scientists from academia and industry to discuss the latest trends and applications in data science, machine learning, and predictive analytics. At our booth, attendees can learn about the benefits of in-memory computing in combating latency issues, how SingleStore enables real-time insights using SQL, and explore our free downloadable Community Edition without limitations on capacity or cluster size. The event also features recommended sessions, including presentations by Comcast and the Berkeley Data Analytics Stack team, which highlight the company's use of SingleStore for real-time stream processing and machine learning pipeline frameworks.

At SingleStore, we see an in-memory, distributed approach to big data as the path forward to cost-effective deployments. Recently, Gartner released a report titled “Five Data Persistence Dilemmas That Will Keep CIOs Up at Night”, which reinforces this approach to data management. The report outlines three key impacts of utilizing new technologies across Hybrid Transaction/Analytical Processing (HTAP) systems that use in-memory processing, the compromises of NoSQL DBMSs, and the growing importance of agile cloud computing approaches. Utilizing HTAP reduces the need for separate dedicated environments and shortens the time to value for new data, but requires process compromises and changes to applications to maximize ROI. NoSQL DBMSs compromise a priori data models and strong levels of consistency to offer high-throughput operations and scale-out architectures. Agile deployment approaches like cloud computing present new opportunities that IT leaders and line-of-business heads must seize. The five dilemmas covered in the report generate questions for CIOs seeking to maximize opportunities with HTAP, scalable SQL databases, and flexible cloud deployments. HTAP collapses the single database model and allows for new classes of applications, while fully realizing its benefits requires thorough cost and capacity planning. In-memory databases like SingleStore offer scalability, performance, simplicity, and SQL. Multi-model databases support structured SQL and semi-structured data types like JSON, providing a balance between structure and flexibility. SingleStore can be deployed on-premises or in the cloud, offering deployment choice. Gartner's report emphasizes the importance of exploring HTAP, scalable SQL, and cloud deployments with in-memory databases like SingleStore.

Join SingleStore on July 15th at 6pm for a meetup in San Francisco's SOMA neighborhood, where database architects and professionals will gather to collaborate and share knowledge. The event features a presentation by Carlos Bueno on the Community Edition and its performance on Amazon's M4 instance, followed by a Q&A session and happy hour. Attendees are encouraged to bring their laptops for hands-on participation. The meetup aims to foster a community of technology professionals in the Bay Area.

** The author of the text designed a benchmark to investigate the performance impact of SSL network encryption on SingleStoreDB Self-Managed 4, a NoSQL database. The test compared the performance with and without SSL enabled, using identical workloads and server configurations. The results showed that enabling SSL resulted in a 25% decrease in write throughput and a 16% drop in read throughput, along with increased median latency. However, the author notes that these findings may not generalize to all scenarios and queries, as the impact of SSL can vary depending on the specific use case. The benchmark serves as an example of how to design and execute a well-structured test to answer specific questions about a database's performance under different conditions.

To build the next great database company, SingleStore seeks high-caliber engineers with expertise in storage systems, code generation, query optimization, and more. The team includes experts from top university programs like MIT, Stanford, and Carnegie Mellon University, as well as highly-rated competition programmers. A growing company with a clear mission to bring database innovation to every enterprise, SingleStore offers a stimulating environment where team members can tackle complex projects creatively and collaborate with colleagues who share similar interests and passions. Engineers at SingleStore value collaboration, impact, and delivering value to customers quickly, and many join the team because of these qualities.