Using Redis at Scale at Twitter – by Rashmi Ramesh of Twitter – RedisConf17 –

Slide deck: https://www.slideshare.net/RedisLabs/redisconf17-using-redis-at-scale-twitter

Timelines, DMs, Twitter Analytics are just some of the major services at Twitter that use Redis. A majority of our analytics services use lambda architecture and they require a highly available cache to store and read stream compute data, while their corresponding batch processing jobs catch up. Our caching solution, built on top of Redis, offers replication for such services requiring a highly available, scalable cache, serving millions of requests per second. In this session, we’ll cover the architecture, constraints and some problems of distributed caching we are trying to tackle today. We’ll also touch upon data partitioning, data movement, failure domain considerations and how we leverage Redis features to achieve the same.