August 2021 Summaries

The blog post compares the performance of Apache Cassandra 4.0 and ScyllaDB 4.4, highlighting significant improvements in ScyllaDB's throughput and administrative efficiency over Cassandra. ScyllaDB demonstrates superior performance with up to 5x greater traffic handling and lower latencies compared to Cassandra 4.0, especially during tasks like node addition, replacement, and major compactions. The benchmarks show that ScyllaDB can achieve similar performance to Cassandra using significantly fewer resources, offering potential cost savings and reduced administrative overhead. While Cassandra 4.0 shows marked improvements over version 3.11, particularly in latency and streaming processes, ScyllaDB provides additional features and efficiencies, such as a more manageable CDC implementation and lightweight transactions. The post suggests that while upgrading from Cassandra 3.11 to 4.0 is beneficial, further gains can be realized by considering ScyllaDB for its enhanced performance and feature set.

Apache Cassandra 4.0 demonstrates significant performance improvements over Cassandra 3.11, primarily due to a major Java Virtual Machine upgrade and the use of the ZGC garbage collector. This upgrade notably enhances latency and throughput, with P99 latencies reduced by up to 99% and throughput increased by up to 33% in some scenarios. Despite these advancements, not all workloads experience substantial gains, as maximum throughput between the two versions remains similar in certain cases. Additionally, Cassandra 4.0 offers improved administrative operation speeds, such as adding nodes and cluster doubling, with up to 34% faster performance. The testing was conducted using typical AWS server setups, focusing on mixed, read-only, and write-only workloads across various data distribution scenarios, including real-life, uniform, and in-memory distributions. However, some users may still face challenges with the transition, considering the stability and maturity of new software features like the ZGC garbage collector.

ScyllaDB Cloud is a Database-as-a-Service (DBaaS) that is gaining popularity due to its ease of use and performance optimization. The platform supports both AWS and Google Cloud, allowing users to choose their preferred cloud provider for hosting clusters with optimized compute instances. Users can also utilize the "Bring Your Own Account" (BYOA) option on AWS to manage their resources and apply any pre-negotiated rates. The service includes a capacity planning calculator to help users size their clusters appropriately based on workload requirements, and deployments are spread across multiple availability zones for high availability. ScyllaDB Cloud includes integrated monitoring for real-time performance visibility, and users are encouraged to conduct stress tests to evaluate the database's performance. The platform supports both its CQL interface and an Amazon DynamoDB-compatible API, while offering robust security features such as VPC peering and strong password protocols.

ScyllaDB University recently hosted a successful two-day virtual training event, the ScyllaDB University LIVE Summer Session, which saw a 27% increase in attendance from its previous session. Participants from major companies like Amazon, Disney, and Salesforce engaged in sessions and an expert panel that covered topics ranging from ScyllaDB's features and benefits over Cassandra, to data modeling and application architecture strategies. Attendees were particularly interested in deploying ScyllaDB Open Source, with the majority managing data workloads in the low-terabyte range. The event also highlighted ScyllaDB's offerings such as its Materialized Views, Change Data Capture, and Lightweight Transactions, which are noted for their enhanced performance and efficiency compared to similar features in Cassandra. Additionally, advanced sessions addressed the integration of ScyllaDB with tools like Apache Kafka and Spark, as well as monitoring techniques to optimize application performance. Despite most attendees being new to ScyllaDB, the event provided valuable insights and resources, with opportunities to continue learning through ScyllaDB University's free, self-paced online courses.

ScyllaDB has announced the general availability of ScyllaDB Cloud on Google Cloud, extending its fully managed NoSQL database-as-a-service (DBaaS) beyond its initial launch on AWS. Designed for modern cloud environments, ScyllaDB Cloud supports both Apache Cassandra CQL and DynamoDB APIs, making it the first managed database offering that allows DynamoDB workloads on a public cloud other than AWS. The platform is recognized for its high performance and cost-effectiveness, with features such as automatic data replication, single-tenant security, and automated backups. ScyllaDB Cloud's deployment on Google Cloud's n2-highmem servers promises enhanced throughput and low-latency performance, enabling organizations like Zeotap and Investing.com to significantly improve data processing speeds and reduce operational costs. The partnership with SADA Systems further facilitates seamless integration with Google Cloud services, providing users flexibility in managing ScyllaDB workloads across different cloud infrastructures. The service offers scalable pricing options, including annual and on-demand billing, supported by a cloud pricing calculator to tailor costs according to user needs.

Cassandra 4.0's release marks a significant, albeit slow, evolution in the wide column store database landscape, maintaining its industry influence despite a notable drop in development momentum and contributions primarily from Apple and Datastax. While the latest version offers improved stability, with more than 1,000 bug fixes and a 25% performance boost due to a new JVM and garbage collection algorithm, its pace of innovation lags behind competitors like ScyllaDB. ScyllaDB, inspired by Cassandra yet reimplemented in C++ for enhanced performance, positions itself as a superior alternative with features such as comprehensive Change Data Capture, scalable architecture, and support for various APIs, including CQL and DynamoDB-compatible options. Emphasizing its open-source roots and active community contributions, ScyllaDB aims to surpass Cassandra by integrating advanced technologies like the Raft consensus protocol to enhance consistency and scalability. Despite Cassandra's continued presence and widespread usage, ScyllaDB argues for its potential to lead the future of wide column store databases, attracting a diverse range of users and developers while encouraging the evolution of the broader ecosystem through shared CQL-based tools and innovations.