Elasticsearch training

This Elasticsearch training is the foundation for all the others: it covers the Lucene-based search and analytics engine in depth, from the user's point of view — storing documents, modeling them, searching them with relevance and getting statistics out of them.

The course is up to date with Elasticsearch 8.x and 9.x: on top of the fundamentals (mapping, text analysis, Query DSL, aggregations), it covers ES|QL and vector, semantic and hybrid search. Every topic is built on a real dataset and a lab, so your teams can integrate Elasticsearch into their projects on their own.

Learning objectives

• Understand how Elasticsearch and Lucene work: inverted index, shards, near real-time.
• Index, update and manipulate documents, one by one or in bulk.
• Model data: mapping, field types, relationships between documents.
• Master text analysis for relevant search, in any language.
• Write Query DSL queries and tune result relevance.
• Compute statistics with aggregations and ES|QL.
• Implement vector, semantic and hybrid search.
• Build application search features: autocomplete, suggestions, geo, highlighting.

Module 1 — Fundamentals

• History, licensing and use cases: search engine, analytics, logs, vector database.
• Lucene and the inverted index; Elasticsearch as a distributed layer and REST API.
• Indices, documents, shards and replicas: what users need to know; near real-time and refresh.
• Local quick start (Docker), Kibana Dev Tools, official clients (Python, Java, JavaScript…).
• Lab: first index and first queries.

Module 2 — Indexing and managing documents

• Index, get, update (partial, upsert, script), delete.
• Versioning and optimistic concurrency (if_seq_no / if_primary_term).
• Bulk API: mass loading and error handling.
• Update by query, delete by query, reindex.
• Lab: loading a real dataset.

Module 3 — Data modeling and mapping

• Field types: text, keyword, numeric, date, boolean, geo_point, flattened, dense_vector…
• Dynamic mapping, dynamic templates and explicit mapping; multi-fields.
• Evolving a mapping: what can change, what requires a reindex.
• Relationships: denormalization, nested, parent / child (join) — benefits and costs.
• Lab: designing a product catalog mapping.

Module 4 — Text analysis

• How text is actually stored: analyzers, character filters, tokenizers, token filters; Analyze API.
• Language analyzers: stemming, elision, asciifolding, stop words.
• Synonyms (synonyms API), shingles, ngrams and edge ngrams, normalizers.
• Index-time vs search-time analysis.
• Lab: a custom analyzer for multilingual content.

Module 5 — Searching with the Query DSL

• Query context vs filter context.
• Full-text queries: match, match_phrase, multi_match, query_string, simple_query_string.
• Term-level queries: term, terms, range, exists, prefix, wildcard, fuzzy.
• Combining with bool: must, should, filter, must_not.
• Sorting, pagination (from / size, search_after + point in time), _source filtering, collapse.
• Lab: a multi-criteria search engine.

Module 6 — Relevance and scoring

• Scoring in Lucene: from TF-IDF to BM25; Explain API.
• Boosts, function_score, script_score, rank_feature, decay functions.
• Rescoring and reranking.
• Measuring relevance: Ranking Evaluation API.
• Lab: improving result ranking.

Module 7 — Aggregations

• Metrics: avg, sum, min, max, stats, cardinality, percentiles (approximate statistics).
• Buckets: terms, range, histogram, date_histogram, filters.
• Nesting, pipeline aggregations, aggregations on nested documents.
• Search facets, composite aggregation for paging.
• Lab: an analytics dashboard through the API.

Module 8 — ES|QL

• Elasticsearch's piped query language: FROM, WHERE, EVAL, STATS, SORT, KEEP.
• Enrichment and joins: ENRICH, LOOKUP JOIN.
• ES|QL vs Query DSL: when to use which.
• Lab: ad hoc analysis with ES|QL.

Module 9 — Vector, semantic and hybrid search

• Embeddings and similarity; dense_vector field, kNN search (HNSW), quantization.
• Semantic search: semantic_text, ELSER (sparse), Inference API and external models.
• Hybrid search: lexical + vector, retrievers, Reciprocal Rank Fusion (RRF).
• Elasticsearch as a RAG building block for generative AI applications.
• Lab: hybrid search over a document corpus.

Module 10 — Application search features

• Autocomplete: completion suggester, search_as_you_type.
• "Did you mean": term and phrase suggesters, fuzziness.
• Result highlighting.
• Geo data: geo_distance, geo_bounding_box, sorting by distance.
• Search templates, percolator, zero-downtime reindex with aliases.
• Lab: a complete search bar.

Going further: running a cluster in production is covered in the Elasticsearch administration training, data collection in the ELK / Elastic Stack training and visualization in the Kibana training.

With over 160 training sessions conducted on Elastic technologies, your instructor is engaged in production work for 50% of the time, serving as an Elk and Elastic Stack consultant . You have an instructor who also possesses hands-on production experience.

Learn more about your instructor.