December 2019 Summaries

MistAway, a leading company in outdoor insect control misting systems, transitioned to ScyllaDB Cloud to enhance its Internet of Things (IoT) infrastructure for managing mosquito populations. Initially relying on MySQL and a third-party SQL database, MistAway faced challenges when their vendor shifted business models, prompting a migration to ScyllaDB Cloud. This move allowed MistAway to efficiently manage over 3,000 systems by leveraging ScyllaDB's high-performance NoSQL database, enabling real-time data analysis and seamless system monitoring. The migration was described as "zero stress" due to ScyllaDB's robust support and comprehensive documentation, ensuring no downtime and minimal maintenance. This transition facilitated improved service delivery by allowing MistAway to precisely address system issues, reducing unnecessary service calls and enhancing customer satisfaction. The switch to ScyllaDB empowered MistAway to focus on value creation and customer service, relieving them from the burdens of database management and infrastructure scaling, ultimately leading to faster data queries and improved operational efficiency.

SkyElectric is addressing Pakistan's energy challenges by implementing a Smart Energy System that combines solar hybrid inverters, grid interconnects, and lithium-ion batteries to provide a reliable power supply, particularly in rural areas affected by frequent power cuts. This system is managed through a cloud platform utilizing ScyllaDB for efficient data handling, enabling real-time monitoring and AI-driven energy management via their SmartFlow solution. Initially, the backend faced scalability issues, which were resolved by transitioning to a robust architecture employing ScyllaDB, PostgreSQL, and Elasticsearch, enhancing performance and scalability. The shift to ScyllaDB, chosen for its C++ foundation and superior resource utilization, has allowed SkyElectric to manage vast amounts of data seamlessly, reducing latencies and enabling the company to expand its operations confidently. The system's efficient management ensures high power availability at low costs, contributing to reducing CO2 emissions and supporting Pakistan's goal to increase renewable energy usage significantly by 2030.

In a blog post detailing the performance capabilities of ScyllaDB, a highly scalable NoSQL database, the author describes a test conducted to measure its speed on Packet's bare metal cloud infrastructure. The test showcased ScyllaDB's ability to scan over one billion rows per second using just 83 nodes, highlighting its efficiency compared to previous benchmarks by other database solutions. The blog also provides a historical overview of NoSQL scalability milestones, emphasizing the importance of understanding the specific metrics and contexts of such performance claims. The test involved simulating a scenario with one million smart home sensors generating temperature data, demonstrating ScyllaDB's capability to efficiently process vast amounts of time-series data. The infrastructure used for the test included high-performance servers provided by Packet, ensuring that network IO was not a bottleneck. The post concludes by encouraging users to share their own scalability achievements and introduces new features in ScyllaDB that enhance its utility in handling large-scale data processing tasks.

Project Gemini is an open-source automated random testing suite designed for ScyllaDB and Cassandra clusters, aimed at ensuring data integrity and reliability in distributed databases. Developed by ScyllaDB engineers, Gemini applies random CQL operations to both a system under test and a reference test oracle to verify that both systems are in the same state, highlighting discrepancies as potential bugs. This tool is particularly useful for quality assurance teams testing CQL-compliant databases, as it aids in detecting hard-to-trigger bugs that can lead to data corruption or loss. By employing automatic random testing, Gemini effectively discovers edge cases, such as reactor stalls and out-of-memory errors, which are not typically covered by other test suites. The tool is implemented in Go, utilizing the gocql driver for CQL clusters, and provides flexibility in data generation and validation processes to simulate various operational scenarios within databases. The project is available on GitHub, offering resources for quick setup and further exploration of Gemini's architecture.

Arm processors, traditionally dominant in the mobile sector, are poised for significant growth in the server market, challenging the longstanding dominance of Intel and AMD's x86 architecture. This potential shift is underscored by AWS's release of the Graviton2-based chips, part of their EC2 M6g and M6gd instances, which demonstrate competitive performance and economic advantages over x86 counterparts. Despite initial underwhelming results with AWS's earlier EC2 A1 instances, the improved performance of the M6 series, particularly in energy efficiency and integration with fast NVMe storage, suggests Arm's capability to transform server economics, especially within cloud-native environments. This evolution is further supported by startups like Nuvia, aiming to standardize Arm-based platforms and capitalize on changes in cloud computing dynamics, where hardware selection is increasingly influenced by cloud providers rather than individual organizations. As cloud gaming and applications become more mainstream, the hardware lock-in diminishes, allowing economic factors to drive platform choices, positioning Arm as a viable alternative in a market ripe for change.

Amazon's new Managed Cassandra Service (MCS) on AWS has sparked interest in the tech community, particularly among those familiar with NoSQL databases like ScyllaDB and DynamoDB. MCS serves as a hybrid, combining the front end of Apache Cassandra with a back end powered by DynamoDB, offering a serverless experience but with certain limitations, such as lack of multi-region support and restricted consistency levels. Despite its potential benefits, including integration with AWS IAM and a convenient serverless setup, MCS faces critiques for its cost and functional constraints, making it more suitable for specific use cases, such as pure Cassandra migrations. The service has been described as a 'chimera'—part Cassandra, part Dynamo—and while AWS promises contributions to the open-source community, skepticism remains about whether these will extend beyond proprietary interests. ScyllaDB, a competitor, views the situation as an opportunity to learn and improve its cloud offering, ScyllaDB Cloud, which boasts significant price and performance advantages over other NoSQL databases.

ScyllaDB Open Source 3.1.2 is a bugfix release that ensures backward compatibility and supports rolling upgrades for the ScyllaDB 3.1 stable branch. This release addresses several critical issues, including incorrect key types for non-frozen map virtual columns, TTLed columns returning null values shortly before expiration, and incorrect aggregated results during queries with paging and filtering. It also resolves stability concerns such as non-graceful handling of disk space limits, crashes during manual operations, and various race conditions that could lead to crashes or incorrect node behavior. For users upgrading from a fresh ScyllaDB 3.1.0 installation, enabling compatibility mode in the configuration file is necessary. Additionally, the release improves user experience by fixing errors in warning messages and issues with the setup script. Users are encouraged to report any encountered problems through the user mailing list.

ScyllaDB Manager 1.4.3, released by the ScyllaDB team, is a bug-fix update to the existing ScyllaDB Manager 1.4, designed to automate maintenance tasks for ScyllaDB clusters. This release addresses issues such as improper error messaging when the certificates directory was inaccessible and incorrect progress values shown by the sctool task progress command during repair jobs. Additionally, the update introduces support for the mTLS protocol in the ScyllaDB Manager REST API, enhancing security features. ScyllaDB Manager is available for both ScyllaDB Enterprise customers and open-source users, with a limitation of five nodes for the latter.

ScyllaDB Enterprise 2019.1.3 is a production-ready patch release focused on enhancing the stability and security of the high-performance NoSQL database designed for real-time big data workloads. This update introduces improvements in encryption at rest by adding two new key providers: the Table provider, which stores table keys within ScyllaDB Tables, and the KMIP provider, which enables secure key exchange using any KMIP-compatible server. The release addresses several stability issues, such as schema alteration handling, cache accounting, node boot race conditions, and performance-related bugs like incorrect workload prioritization. Additionally, it includes fixes for hinted handoff races, commit log errors, and Docker command-line parsing issues. ScyllaDB Enterprise customers are encouraged to coordinate with the ScyllaDB support team for upgrading to this version.