August 2014 Summaries

At the Game Developers Conference (GDC) and other events like GDC Europe and Gamescom, game developers use Redis for various purposes including leaderboards, sessions, profiles, caching, and real-time data management. Online, social, and mobile games require strict latency and performance requirements, making Redis a suitable choice. Ad networks also heavily rely on Redis due to its speed and ability to handle high volumes of requests. The flexibility of Redis is appealing to developers, who often implement custom solutions using Lua scripts or modify the existing code. This versatility has led to creative uses in game development, such as implementing player matching mechanisms.

Tung Nguyen will be speaking at the NoSQL Now! 2014 event on August 19th, discussing how Bleacher Report utilized Redis to create a highly scalable, available, and top-performing application with sub 60ms response times. Redis is an open source in-memory NoSQL database known for its sub-millisecond latency. The session will cover the use of Redis by Bleacher Report and how it achieved these impressive results.

Cloud computing can help host an application in a scalable and cost-effective way, particularly with Amazon Web Services (AWS) offering secure cloud hosting with Redis. To optimize performance and reduce latency issues, using HVM-enabled instances can help reduce fork times and run multiple Redis servers on different cores. However, enabling swapping can increase latency due to disk I/O operations, which can be mitigated by configuring Linux settings in `/etc/sysctl.conf`. Additionally, using large EBS volumes, RAID configurations, or delegating persistence can improve performance when persisting data to EBS. It's also essential to review application persistence requirements and tune Redis' persistency accordingly. Furthermore, considering the size of the dataset and provisioning for extra memory requirements can help prevent running out of memory. By efficiently utilizing EC2 instances with multiple cores, starting one Redis process per core or sharding data between servers can improve performance.

I'm a data enthusiast who's fascinated by data management systems and has spent 25 years working with various databases. I'm attending the NoSQLNow! event to learn about new database technologies and dive into datasets, but I'm worried my addiction will get the better of me. The event features top-notch sessions on topics such as modeling for NoSQL and SQL databases, big data engineering, polyglot persistence, and more. With over 30 quality sessions to choose from, I've had to limit myself to just a few highlights, including a case study on Bleacher Report's use of Redis Cloud and a session on big data engineering. The event also features a vendor-independent comparison of NoSQL performance based on real-life cases, and I'm looking forward to learning about in-memory architectures for current workloads and the future of processing.