February 2019 Summaries

Ten years ago, Redis was introduced as a key-value store with complex types and a module system. Its creator Salvatore Sanfilippo (a.k.a. antirez) aimed to support more complex data structures from the beginning. The first post about Redis on Hacker News was shared by David Welton (a.k.a. davidw), followed by Ezra Zygmuntowicz (a.k.a. ezmobius, RIP) who encouraged Salvatore and implemented the first Ruby client library for Redis. This helped spread Redis among Ruby on Rails developers, including Twitter, and eventually made it a popular add-on for Heroku developers. The early adopters of Redis played an essential role in its growth as an open source community.

Redis has changed its license for certain Redis Modules to the Redis Source Available License (RSAL), addressing concerns about cloud providers using successful open source projects without contributing back. RSAL maintains equivalent rights to permissive open source licenses, with a few restrictions on specific types of applications. This change does not affect the Redis core license, which remains under 3-Clause BSD. The company continues to focus on providing instant experiences for modern applications through its open approach and commitment to maintaining an independent Redis core team.

The data landscape has grown increasingly demanding and crowded in the last few years, with many platforms competing to offer the best processing and storage options. Tech consumers expect companies to deliver high-speed data processing while simultaneously offering a variety of adaptable solutions that work with traditional applications but also built for modern architecture like microservices. A monolithic framework guarantees data consistency, which is important when adopting a microservice architecture. Redis Enterprise offers two options for data synchronization: sharing datasets across microservices and transferring data between them. The former relies on conflict-free replicated data types, ensuring strong eventual consistency, while the latter employs Pub/Sub, Lists, Sorted Sets or Streams to facilitate event-based architectures. Redis Enterprise is highly available, durable, multi-tenant and Kubernetes supported, with cloud and on-premises options, making it an ideal database solution for microservice architecture.

Apache Spark is a popular framework for creating distributed data processing pipelines. It allows for flexibility in running pipelines both locally and on a cluster without changing source code. The framework uses delayed computation or laziness to achieve this. Spark has three APIs for dealing with distributed datasets, but each API builds upon the previous one. In this article, we explored how to use Redis as a backend for Spark DataFrames using Python, specifically focusing on getting started with the DataFrame API and performing common operations such as filtering data based on occupation and country. The process involves installing pyspark, building spark-redis, setting up a Redis server, loading data into Redis, and writing a pipeline to get the most frequent occupation for famous people in each country. We also discussed the importance of scaling Redis appropriately using the Redis Cluster API to avoid bottlenecks when dealing with large datasets.

The author emphasizes the importance of breaking down applications into smaller, loosely coupled microservices that work together. They discuss the challenges of inter-services communication models and highlight the need for a robust system to manage state and events between services. The author recently adopted Redis Streams as a solution for managing microservices communication at scale, citing its benefits such as easy implementation of pub-sub patterns, support for time-series data, Active-Active capabilities, simple deployment, and in-memory processing. They demonstrate how to create an event store using Redis Streams by building an e-commerce application with multiple microservices that use the system to communicate events asynchronously. The author uses Redis Streams to model their domain model, separating commands from queries and utilizing Sets to store entity IDs and Hashes to represent data structures. They highlight the flexibility of Redis data structures and their ability to adapt to various data models, making it an attractive choice for enterprise-ready applications.

As a Customer Advocacy Manager at Redis Labs, I have the privilege of hearing first-hand accounts from customers who use our product, including their successes and challenges. The success of my role is heavily dependent on customer happiness, as everyone's job at Redis Labs is tied to customer satisfaction. When we were named a Gartner Peer Insights Customers' Choice in 2019, it was a testament to the support of an active and passionate community that helped us maintain a 4.6 rating based on over 100 reviews. This recognition highlights the importance of customer feedback in shaping our products and journey, and we look forward to continuing to foster these relationships with our customers who submitted reviews. Our goal is to build on the experiences that contributed to this distinction, and we encourage others to share their Redis stories by joining Gartner Peer Insights.