Search
ElasticSearch from the bottom up
I really love this talk as Alex taking us from the high level cluster all the way down to the inverted index in Lucene. Although some of the interesting topics like finite state transducers, inverted index are just skimmed through, it still gives me a great visual picture of how things work behind the scene.
Speed is Key
It is fascinating to find out how ElasticSearch/ Lucene achieve the great speed. Some of the topics like HyperLogLog++ for counting distinct values and T-Digest for percentile calculation worths another articles to talka about.
Data Design Pattern
To build a cluster and index 100K website content and support all their search traffics requires a well thought of how we can use ElasticSearch efficiently. The video below shows me how we can combine routing and alias smartly to make it happened.