December 2025 Summaries

In 2025, ScyllaDB's blog posts covered a diverse range of topics, from technical advancements to practical applications in big data and cloud computing, reflecting the community's ongoing interest in database performance and scalability. Notable posts included insights into the ScyllaDB Rust Driver 1.0, which introduced an async shard-aware CQL driver using Tokio, and a technical overview of ScyllaDB X Cloud, highlighting its elasticity and low-latency support for variable workloads. The blogs also explored real-time personalization at Tripadvisor using ScyllaDB and AWS, Supercell's handling of real-time persisted events, and the transition from mutation-based to file-based data streaming for faster performance. Other discussions addressed the reasons some teams are moving away from DynamoDB for lower latency and operational costs, the implementation of ScyllaDB Bloom filters by JioCinema for personalization, and the optimization of full-table scans using ScyllaDB Tablets. Additionally, the blog revisited classic posts that continued to engage readers, such as benchmarks comparing ScyllaDB with other databases and debates on systems programming languages like Rust and Zig.

ScyllaDB Operator 1.19.0 introduces several significant enhancements, including multi-tenant monitoring with Prometheus and OpenShift support, an improved sysctl configuration API, and an opt-in synchronization feature for safer topology changes. This release offers "External mode," which allows multiple ScyllaDB clusters to share an existing Prometheus deployment for monitoring, facilitating integration with Red Hat OpenShift's User Workload Monitoring. Additionally, the must-gather tool now includes a guardrail to prevent the accidental inclusion of sensitive information, while sysctl configurations are optimized by moving them to a more appropriate API location. The update also introduces a barrier to ensure no nodes are down during topology changes, though this is an opt-in feature. Deprecated features include the previous sysctl configuration method and exposeOptions in ScyllaDBMonitoring components, with users encouraged to transition to more flexible alternatives. The release also aligns with updated dependencies and lays the groundwork for future deprecations of the "Managed mode" monitoring setup.

The blog post introduces a new guide on using the ScyllaDB Change Data Capture (CDC) with the Rust programming language, adding to previous resources for Java and Go. It highlights the scylla-cdc-rust library, which simplifies the process of consuming CDC logs in Rust applications by handling errors and topology changes of the ScyllaDB cluster. The guide walks through setting up a Rust application to read and print real-time changes to a database table using this library. It emphasizes the importance of defining a callback function, implementing relevant Rust traits for asynchronous operations, and managing shared state between consumer instances efficiently. The post also provides a brief overview of configuring the application, including establishing a database connection and setting up a log reader. Additionally, it suggests further reading materials for more complex implementations in Java and Go, offering detailed examples and documentation links for those interested in exploring CDC capabilities further.

At the recent re:Invent conference in Las Vegas, ScyllaDB made a significant impact with its largest presence to date, featuring a full-scale booth and numerous presentations showcasing its capabilities in handling extreme scale database operations. The event, attended by 65,000 tech enthusiasts, highlighted ScyllaDB's ability to manage millions of operations per second with minimal latency, demonstrating its effectiveness in real-time AI workloads and vector search applications. The booth buzzed with activity as data engineers, developers, and decision-makers explored ScyllaDB's seamless integration with existing Cassandra or DynamoDB environments, its cost-effective performance, and its flexibility across different cloud platforms. The conference underscored the growing importance of real-time systems in database engineering, emphasizing ScyllaDB's role in providing scalable, low-latency solutions for AI inference and IoT telemetry.

ScyllaDB Vector Search is a high-performance solution designed to handle billion-scale datasets with ultra-low latency and high throughput, as validated by a benchmark using the yandex-deep_1b dataset containing 1 billion vectors of 96 dimensions. The system achieves this through an architecture that separates storage and indexing duties while maintaining a unified user perspective, with nodes storing structured data and vector embeddings in a distributed table. A dedicated Vector Store service, implemented in Rust and powered by the USearch engine, builds approximate-nearest-neighbour indexes in memory to ensure predictable single-digit millisecond latencies. Two usage scenarios were tested: one prioritized ultra-low latency with moderate recall, achieving 252,000 queries per second, while the other focused on high recall with slightly higher latency, maintaining 6,500 queries per second. ScyllaDB integrates structured and unstructured data retrieval, simplifying operational complexity by eliminating the need for separate systems and reducing network costs. With planned enhancements, including scalar quantization and sharding, ScyllaDB aims to further boost performance for real-time AI applications, offering a scalable and reliable solution for latency-critical tasks such as fraud detection and recommendation systems.