Skip to content

ElasticSearch

ElasticSearch#

- Shay Banon, 2010

Table of Contents

What#

  • ElasticSearch (ES) is a search and analytic engine (more than a No-SQL database)
  • based on Apache Lucene library
  • provides
    • distributed
    • multi-tenant capable
    • full-text search engine
  • through
    • HTTP web interface
    • and (schema-free) JSON documents
  • kind: index & document based

How#

How does ElasticSearch work?

  • ES uses a concept/data structure called Inverted Index
    • which is designed to perform very fast full-text search

  • A document (says string/sentence) saved as well as is analyzed and tokenized for better search results

    • the sentance, say "this is a test string" is split into words, n-grams
    • and each part is mapped to relevant document id, position, and number of occurrences

  • ES index is made up of one or more shards

  • each shard can have zero or more replicas
    • these are individual Lucene indices
  • one ES index is made up of multiple Lucene indices
  • either a single index's multiple shards could be searched, even multiple ES indices could be searched for a term
    • any of this is just combination of multiple Lucene indices
  • shard is the main scaling unit in ES

Components#

  • Nodes
    • Master
      • Controls the Elasticsearch cluster and is responsible for all cluster-wide operations like creating/deleting an index and adding/removing nodes.
    • Data
      • Stores data and executes data-related operations such as search and aggregation.
    • Client
      • Forwards cluster requests to the master node and data-related requests to data nodes.
  • Indices
    • Index
      • Shard
        • Document
  • Replicas
    • of shards

Ref: - https://www.elastic.co/blog/found-elasticsearch-from-the-bottom-up - https://www.elastic.co/what-is/elasticsearch - https://www.knowi.com/blog/what-is-elastic-search/