ElasticSearch Overview

Transcript

1. ElasticSearch Srinivas Singanamalla
2. 2 ElasticSearch Search Analytics Schema-less JSON Engine Java based Lucene Opensource* Near Real Time Distributed Scalable
3. 3 Google Trends
5. 5 Jim Elasticsearchovsky Faceboook Inc. Alice Faceboookuserger Faceboook User No intent of gender discrimination
6. 6 Jim Elasticsearchovsky Faceboook Inc. Alice Faceboookuserger Faceboook User
8. 8 Faceboook Post
9. 9 Full Text Search • Query – message: “color pink” • Results – “Why do girls have to like color pink ?” – “Pink and blue arrived as color for babies in the mid-19th century” – “I love to color blue” – “This blue color is amazing”
10. 10 Phrase Search • Query – message: “blue color” • Results – “This blue color is amazing”
11. 11 Highlighting Search • Query – message: “elasticsearch” • Results
12. 12 Social Likeability: The number of likes you get in your social media content.
13. 13 Numeric Search • Query – likes: 50 • Results – None
14. 14 Numeric Range Search • Query – likes > 50 • Results – “Let us get together and drain that swamp tomorrow” – “How could the polls be so wrong?”
15. 15 Call me? 312 4?9 2868
16. 16 Fuzzy Search • Query – message: “312492868” • Results – “It is not good to share phone numbers, but here it is: 6312492868” – “Utility Bill number: 3124928685” – “Bel me 3124892868” 312 4?9 2868
18. 18 Search Suggestions • Query – “tring out Elasticsearch” • Results – “trying out Elasticsearch”
19. 19 Other Searches • Search Boosting – author:douglas OR title:guide^5 • Proximity Search – “douglas guide” • Regular Expressions
20. 20 Aggregations • Count • Sum • Average • Max/Min • Median • Standard Deviation • Percentile
21. 21 Search Query: REST API
23. 23 Aggregation: Sum
24. 24 Aggregation: Average
25. 25 DEMO
26. 26 Search and Aggregations Ingestion Inverted Index Doc Values Search Aggregation
27. 27 Inverted Index
28. 28 Inverted Index Table Term Document Id Frequency i 1 1 love 1 1 to 1 1 color 1, 2 2 blue 1, 2 2 sky 2 1 amazing 2 1
29. 29 Inverted Index: Numeric Search • Find documents where – 423 < clicks < 642
30. 30 Aggregations: Sum Get me sum of all likes
31. 31 Doc Values (Column Oriented) Doc Id Value 1 10 2 5 3 120 Likes Likes: [10, 5, 120] Getting all the likes is super fast Sum: 10 + 5 + 120 = 135
32. 32 Filtered Sum Get me sum of all likes containing “elasticsearch”
33. 33 Filtered Sum • “Likes” field values: [10, 5, 120] • query_bits: [0, 1, 1] • Apply_query: [0, 5, 120] • Calculate Sum 0 + 5 + 120 = 125
34. 34 How can we use ES? API Server RDBMS ElasticSearch
35. 35 How can we use ES?
36. 36 Finally Ask not what ElasticSearch can do for you -- ask what you can do with ElasticSearch.
37. 37 • Icons are taken from http://www.freepik.com/