August 2018 Summaries

Hazelcast has published two benchmarks comparing its system against the Redis open source database. In response, a similar benchmark was run between Redis Enterprise and Hazelcast using AWS servers with Ubuntu 16.04 OS version and cluster versions of 5.2.0 for Redis Enterprise and 3.9 for Hazelcast. The benchmark used memtier_benchmark version 1.2.8 for Redis Enterprise and RadarGun version 3.0.0 for Hazelcast, with a read:write ratio of 80:20, 4 million objects, object sizes ranging from 100 bytes to 10,000 bytes, and a total dataset size of 42GB. The results showed that the Redis Enterprise cluster had a much better throughput (over 3.5X) and latency (~3X) compared to Hazelcast.

The text discusses the deployment of Redis Enterprise on Kubernetes using a Pivotal Container Service® (PKS) cluster. It explains why PKS is useful for managing containerized microservices in a cloud-native way and how it can be combined with the Pivotal Application System® to manage an entire application lifecycle. The text also outlines the principles used for deploying Redis Enterprise on PKS, including operator-based deployment, network-attached persistent storage, layered orchestrator architecture, and multiple instance deployment. It provides instructions on how to get started with Redis Enterprise on PKS and benchmark its performance. Finally, it mentions future integration efforts around areas of network segmentation, Kubernetes ingress primitive, and support for Active-Active geo-distributed Redis CRDTs over PKS.

RediSearch 1.4 introduces two key features that enhance querying capabilities: spell check and phonetic matching. Spell check uses a primitive to power a "did-you-mean" feature, identifying misspelled words and suggesting corrections. Phonetic matching solves the canonical problem of searching for similar names by using an algorithm called Double Metaphone, which breaks down text into language-specific codes based on pronunciation rules. These features add flexibility to search while accommodating human error without losing sight of user intention.

The recent licensing change for Redis modules has led to confusion about its implications, with some misinformation circulating. The open source Redis license remains unchanged, still being BSD. However, the license for Redis modules developed by Redis was changed from AGPL to Apache v2.0 modified with Commons Clause. This change aims to allow full use of the modules under a liberal license while restricting direct sales of the original modules and promoting the community's interests over cloud providers' profits. By adopting this new license, Redis seeks to maintain its rights over commercializing assets while supporting open and free use of the modules. The move was made in response to the fact that cloud providers have been profiting from open source code without contributing to its development.

For three years running, Redis Day Tel Aviv has provided a platform for developers to share their stories about using Redis. This year, an expansion to include Redis Day London is taking place, featuring a diverse lineup of speakers from the community. Speakers will be discussing various topics related to Redis, including its use in high-traffic applications and new features like RESP3 protocol and client-side caching. The event aims to bring together developers, end-users, and contributors to Redis for an informative and engaging experience.

Redis Enterprise 5.0 introduces support for the Open Source cluster API, allowing Redis clusters to scale infinitely by adding shards and nodes, with linear performance scalability demonstrated through benchmark tests that show sub-millisecond latency across all data sizes and workloads as throughput increases. The database can scale both out and up, optimizing for both linear scaling by adding resources in a deterministic manner and true linear scalability, which translates to predictable performance with efficient resource use, helping build scalable modern applications cost-effectively.