Yomali

We use Node.js because of its speed, efficiency and elegance. Node.js event-driven architecture makes it possible to use a fraction of the servers we needed before, delivering better performance and all with much fewer lines of code. The rich ecosystem and common libraries make Node.js development a heavenly experience.

We use Elasticsearch because of its speed and elasticity. Used pure Lucene in the past. ElasticSearch adds a rich query language, facets, replication and semi-automatic recovery. We're using ElasticSearch in a multi-datacenter master-master environment, but push the writes to each datacenter using a home-grown jobqueue.

We use Redis because of its speed and flexibility. When you need to do analytics, caching, pubsub and locking at scale, at sub second response times, we didn't find any engine that beats Redis. Found the Redis Cluster solution unreliable, so we replaced it with home-grown master-master partitioning and replication of our data.

We use MySQL because if you shard your data correctly, nothing beats the simplicity, ease of maintenance and reliability of MySQL + InnoDB. We are using several additional database storage engines in our stack, but MySQL is by far the most reliable and battle tested.