April 2019 Summaries

The author of the text started building websites without focusing on performance, assuming it was a luxury for later stages of development. However, they soon realized the importance of caching and its impact on SEO, user experience, and revenue. The author implemented a basic page caching system and learned that while it worked well, it was only one tool in a larger solution. They explored how to cache database queries using Redis, which can significantly speed up their site by reducing the time spent accessing data from the database. By storing cached query results in Redis, they can reduce the number of database queries and improve performance, leading to faster page loads and increased user engagement.

Redis can significantly enhance your continuous integration and continuous deployment (CI/CD) process by providing tools for feature toggling, feature context, error logs, and centralized data storage. Feature toggles allow you to rapidly deliver new functionality while maintaining control over its release, whereas feature context enables dynamic routing decisions based on user context such as session management. An error database helps with error triage by storing errors reported during runtime in a powerful text search engine called RediSearch. Additionally, a logs database provides a centralized data store for tracking log messages, allowing you to easily access and manage your application's recent log history. By leveraging Redis Enterprise and its features such as data persistence and secondary indexing, developers can simplify their CI/CD process and minimize headaches associated with managing large-scale apps with frequent releases.

Redis Day New York is set to take place on June 27th and will be the first event of its kind in the city, following successful editions in Tel Aviv and London. Unlike RedisConf, Redis Day features a single track with short talks and no public Q&A per session. The organizers are currently seeking talk submissions that focus on Redis and its applications. Early submission is encouraged as it increases the chances of acceptance due to the limited number of slots available. To submit a talk, visit the Redis Day New York CfP page and fill out the form.

In an effort to run smoothly, it's crucial to learn from past mistakes like Knights Capital's 2012 system failure that cost $460 million. This experience highlights the importance of designing, implementing, and optimizing system components, especially for continuous integration and deployment processes. A well-managed CI/CD process can help development teams manage updates efficiently by using feature toggles and feature context to dictate code routing, storing log data for easy access, and creating an error database with fast lookups - all supported by Redis.

RediSearch is a high-performance search engine built as a Redis module, enabling fast execution of complex search queries on large datasets. It was benchmarked against Elasticsearch and found to be 58% faster in indexing and 4 times faster in throughput, with slightly better latency. RediSearch's optimized data structures and native Redis protocol make it a viable alternative to other search engines. The engine's performance advantages are attributed to its C-based architecture, native memory technology, and dedicated search engine design.

Inovonics, a leading provider of industrial wireless IoT technology, has transitioned its data management from on-premises desktop applications to a cloud-based system utilizing Redis Enterprise Cloud Pro on Google Cloud Platform. This shift enables the company to consolidate and analyze vast amounts of data collected by its devices and sensors, providing insights for predictive data analysis and aggregations in various markets such as life safety and submetering. The high reliability and performance of Redis Enterprise have allowed Inovonics to focus on building end-user solutions rather than optimizing database access, ultimately enhancing the value it provides to its customers.

We just released a new version of Redis Enterprise – v5.4.2 and are happy to announce that we’ve added improved compatibility for active-active Redis (through CRDBs, or conflict-free replicated databases) COUNTER operations with open source Redis. Prior to this release, COUNTER operations were treated differently in Redis due to the nature of CRDBs, which created conflicts with users' expectations. The new version resolves these issues by treating COUNTER operations uniformly across both Redis and CRDBs. This is a significant improvement for applications that use CRDBs to migrate from standard Redis deployments to active-active deployments. To take advantage of this feature, users need to install the latest Redis Enterprise version 5.4.2 and upgrade their existing CRDB deployments, which will automatically restart the required servers. The new protocol is backward-compatible but not forward-compatible, so it's recommended to upgrade all instances within a reasonable time frame to avoid temporary inconsistencies.

Redis has been named the most loved database in Stack Overflow's 2019 Developer Survey for the third consecutive year, with a significant increase in share among developers as their preferred database, up nearly seven percent to 71.3% of total respondents. This recognition comes directly from the developer community and is seen as a testament to Redis's innovative capabilities and performance. The company continues to innovate and develop Redis to meet the needs of developers worldwide, with industry recognition of its new features. With over 89,000 responses from developers in 179 countries and 1.5 billion launches on Docker Hub, Redis has gained widespread love and adoption by developers globally.

We began our collaboration with Google Cloud in 2014 to make Redis Enterprise available on Google Cloud Platform (GCP) as a fully managed Database-as-a-Service, either hosted or in virtual private clouds (VPCs), as well as downloadable software. Since then our partnership has strengthened considerably as we worked together with many joint customers across every industry including retail, financial services, media, social, and gaming. We share a common vision that open source accelerates innovation and helps our customers future-proof their technology investments. Combining full resilience utilizing Redis Enterprise’s Active-Active Geo-Distribution capability with GCP's fast global network will allow enterprises to deploy distributed applications with sub-millisecond latency. The new partnership will deliver Redis Enterprise as a fully managed cloud service, offering a seamless experience integrated with Google Cloud Platform (GCP) console, billing, and support to enterprise customers.

RedisConf 2019 was a significant conference for Redis users, also known as "Redis Geeks," who gathered at Pier 27 in San Francisco on April 1-3 to share their expertise and knowledge about the world's most-loved database. The event featured over 50 speakers, including keynote speaker Julia Liuson, Corporate Vice President of Microsoft's Developer Division, and included sessions on using Redis in microservices, Time Series Database functionality, and multi-model databases. Notable announcements made during the conference included the release of Redis 6.0 with improved multi-threading and Access Control Lists (ACLs), as well as the introduction of new modules such as RedisTimeSeries and RedisAI. The event also featured partnerships and collaborations with major companies like Google Cloud, Microsoft, and Intel, highlighting the growing ecosystem around Redis Enterprise. Overall, the conference provided a vibrant venue for knowledge exchange among developers and showcased the power of Redis in various industries.

Today at RedisConf 19, Redis announced the acquisition of RDBTools from HashedIn, a leading product for viewing and analyzing Redis data with features like Memory Analysis. The goal is to help developers get the most out of Redis. RDBTools offers rich visualization, insights into Redis deployments, and understanding relationships between Master and Replica instances and Shards. It can be deployed through various means, including desktop clients, Docker images, and a managed service on AWS Marketplace. A free version will be provided with limited features, while a fully-managed service is planned for the future.

Over the past decade, Redis has been used by customers in various ways such as cache, session store, message broker, recommendation engine, secondary index, streaming platform, and single source of truth database. Its ability to process data at sub-millisecond speeds irrespective of data volume makes it extremely versatile across countless scenarios. However, the performance efficiency gained by in-memory computing can sometimes come at a cost, especially if the volume of data needed for instant processing is substantially large and resides in DRAM. To address this issue, Redis introduced Redis on Flash in 2016, which stores keys, data dictionaries, and "hot" (frequently accessed) data in RAM, and "cold" data on Flash SSDs while maintaining sub-millisecond performance. Many customers have achieved significant savings on infrastructure costs with Redis on Flash. Recently, Redis has been working closely with Intel to ensure that their latest memory technology—Intel® Optane™ DC Persistent Memory—is immediately available to Redis Enterprise users upon its general release with the second-generation Intel® Xeon® Scalable platforms. This new tier extends a standard machine's memory capacity to up to 7.5TBs of byte-addressable memory (DRAM + persistent memory), while providing persistence, and is available as a 128, 256, and 512GB persistent memory module. Redis Enterprise users can now keep more data per node, dramatically reducing their infrastructure costs while maintaining performance SLAs, keeping sub-millisecond latency at high throughputs of 1M ops/sec – a typical throughput for Redis Enterprise customers. This results in approximately 43% savings in hardware costs with little to no impact on performance—even on throughputs running into millions of operations per second! Redis has adapted its stack and Redis on Flash to work with Intel's new technology, which is also byte-addressable by design. To learn more about this technology, attendees can join the RedisConf 2019 in San Francisco or download a white paper that delves into detail on the persistent memory technology.

The new RedisAI module for serving tensors and executing deep learning models enables colocality, allowing models to run where the data lives, reducing costs and improving performance. It introduces a new data type called Tensor, Models, and Scripts, which inherit enterprise-grade features of Redis, making it easy to scale model serving and reduce infrastructure costs. With integrated backends such as TensorFlow and PyTorch, RedisAI decouples backend choice from application services, allowing for flexibility and ease of use. The module also includes features like automatic batching, DAG commands, and metrics analysis, which will be added in the near future, enabling users to optimize model performance and efficiency.