October 2022 Summaries

MySQL is a widely used open-source database with companies like Airbnb, Pinterest, and Netflix relying on it for their tech stacks, but its single-node architecture can become limiting when handling real-time analytics on rapidly changing data. SingleStoreDB offers a solution by providing MySQL compatibility, familiar SQL syntax, and features such as Pipelines that enable faster data loading at scale in parallel, making it easier to migrate from traditional databases like AWS RDS, Google Cloud SQL, Azure MySQL, and others. The database also includes JSON support with over 15 functions for building, extracting, deleting, and displaying JSON data, allowing users to easily move their JSON data into SingleStoreDB using the MySQL CLI, SingleStoreDB Pipeline, or the portal UI.

Poor data experiences can be recognized by issues such as application or dashboard failures, outdated supply chain data, inaccurate pricing in ride-sharing apps, and missed fraudulent payments. Today's users demand real-time visibility and insights to make informed decisions, but legacy platforms often struggle to keep up with fast-moving data streams. To deliver on the promise of real-time analytics, databases need to handle streaming data ingestion, low latency analytical queries, flexible indexing, and high concurrency. SingleStoreDB is a solution that meets these requirements with its three-tiered storage architecture, unified architecture, reliable support for complex queries, strong high availability, separation of storage and compute, and various use cases across industries such as adtech, cybersecurity, fintech, and retail.

The development of real-time IoT analytics is crucial for ensuring the health and functionality of various systems, including energy grids and smart meters. This technology enables the identification of critical anomalies and alerts, reducing response times to prevent excessive downtime. A key component in achieving this is ultra-fast data ingestion, low latency, and high concurrency, which SingleStoreDB can provide through its integration with IBM Data Fabric. The platform combines transactional and analytical workloads into a single engine, empowering real-time analytical applications such as IoT analytics, operational BI, ad hoc data discovery, ML-driven analytics, and historical reporting.

We have migrated from Jest and React Testing Library (RTL) to Cypress Component Testing (CCT) for UI integration tests at SingleStore. The main reasons for this migration were the flakiness of our previous integration tests, which were causing issues with debugging and performance. We found that RTL's JSDOM simulated rendering made it difficult to visually inspect test results, leading to slower test execution due to CSS not being considered in tests. Additionally, we experienced issues with Jest's slow startup times as more tests were added. The flakiness of our integration tests was mostly a mystery, but we identified improperly written tests and missing API mocks as common causes. We decided to move to CCT instead of fixing the underlying issues due to its familiarity and the fact that our end-to-end tests were already written with Cypress. Our experience with CCT has been very positive, with improved test realism, faster onboarding for new engineers, and better developer experience through Cypress's API. However, we still have some flakiness in our integration tests that we are investigating.

The company SingleStore has had a successful first half of 2022, closing a $146 million funding round led by Goldman Sachs, expanding its capabilities to address real-time applications, and securing new customers such as tech giants Dell Technologies and Digital Asset Research. Third-party research has shown that SingleStore provides greater value at a lower cost than the competition, earning awards for its value, feature set, and relationship in relational databases. The company is also investing in diversity, equity, and inclusion efforts and is poised to capitalize on growing trends such as real-time analytics, machine learning, and artificial intelligence.