April 2018 Summaries

We are FreeWheel, a Comcast company providing solutions to the Media Advertising industry to enable efficiency and insight into all aspects of campaign workflows, across media for media buyers and publishers. Our system is based on a microservices architecture that has evolved over 15-20 years, necessitating a robust caching mechanism to support shared data storage within our ecosystem and across environments. We initially used a simple per-server-per-application store but faced issues with stale caches and unmaintainable code. After exploring various solutions, we settled on Redis, which brought reliability, performance, and flexibility, along with good community support and documentation. We now deploy Redis Enterprise in production, supporting users globally, and have seen improved system stability and identified new use cases that will enhance the user experience. The transition to Redis has enabled us to design infrastructure for automation of internal processes, facilitating Continuous Integration scenarios.

Redis on Flash (RoF) is a memory technology that takes advantage of high-speed storage in the form of Flash SSDs, resetting price/performance expectations. It extends Redis from RAM to Flash, using intelligent tiering to keep hot values in RAM and cold values in cost-effective flash-based SSDs. This approach reduces RAM consumption and achieves sub-millisecond latencies. RoF has been used by customers such as Whitepages, which reduced its infrastructure costs by hundreds of thousands of dollars per year. Redis is working closely with Intel to ensure its solutions run optimally with new 3D XPoint technology, which delivers a new tier between DRAM and SSD that can provide up to 6TB of capacity in a two-socket server at performance comparable to traditional DRAM memory. This technology allows users to think of memory as the main storage tier for their data and gain the ability to extend their "memory". RoF is not just limited to Redis, but other DBMS must be optimized to work with this new technology.

Redis offers persistence and in-memory replication capabilities, making it suitable for use cases like session management, queues, pub/sub, and general-purpose caching. As a freelancer, Rahul uses Redis to overcome unique problems by leveraging its pub/sub feature to notify processes of new detections, eliminating the need for a daily job. By adopting Redis, he discovered other advantages such as using built-in data structures to perform analytics, gathering information on file changes and applications making those changes. Getting started with Redis is extremely easy, and it can be accessed for free through Redis Cloud's 30 MB plan.

This tutorial uses Kops to set up a production-grade Kubernetes cluster on a public cloud using AWS, and then deploys Redis Enterprise Service on top of it, showcasing the ease of use of Kops in setting up a Kubernetes cluster for a variety of applications. The process involves preparing the local machine with necessary tools, configuring a domain name, creating an S3 bucket to store the state of the cluster configuration, creating a Kubernetes cluster using Kops, deploying Redis Enterprise Service on the cluster, and deleting the cluster when no longer needed.

Redis Enterprise 5.0.2 is now generally available, offering significant enhancements and new features including additional Active-Active capabilities, improved encryption options for data in transit, Redis Enterprise Modules improvements, preview release of Kubernetes support, and integration with Redis Enterprise VPC.