Twitter has hit 1 billion search queries per day. To accommodate the searches and organize tweets, Twitter updated their entire search back-end. Most of the changes were done “under-the-hood.”
Twitter’s real-time search engine was used on original technology based on Summize. Twitter had to scale the old MySQL-based system. Twitter created modern search architecture that is based on an inverted index instead of relational databases. Twitter decided to use a search engine library written in Java called Lucene.
Twitter’s demands on the new system is known as 1,000 TPS (Tweets/sec) and 12,000 QPS (queries/sec). That means that there is over 1 billion queries per day. Twitter rewrote core parts for their in-memory data structures.