Elasticsearch CRUD and Mapping

Index documents, dynamic vs explicit mapping, and keyword vs text fields.

Advanced12 min read

Read these first

Elasticsearch Architecture

Why this matters

VaultCommerce product SKUs must be keyword (exact match, aggregations); descriptions must be text (analyzed for search). Swapping them breaks faceted filters and full-text queries.
Bulk indexing during catalog sync needs the _bulk API — single-document PUTs cannot keep up with nightly imports.
Mapping mistakes (e.g., price mapped as text) require reindex — expensive at millions of documents.

Cluster

Node 1Shard 0 primary

Node 2Shard 1 primary

Node 3Shard 0 replica

An index is split into shards across nodes. Replicas provide failover and read scaling.

Explicit mapping for products

Java

curl -X PUT "localhost:9200/vaultcommerce-products" -H 'Content-Type: application/json' -d'
{
  "mappings": {
    "properties": {
      "sku":        { "type": "keyword" },
      "title":      { "type": "text", "fields": { "raw": { "type": "keyword" } } },
      "description":{ "type": "text" },
      "category":   { "type": "keyword" },
      "price":      { "type": "float" },
      "in_stock":   { "type": "boolean" },
      "tags":       { "type": "keyword" }
    }
  }
}'

keyword vs text

keyword — stored as-is for filters, sorting, aggregations (SKU, category).
text — analyzed (tokenized, lowercased) for full-text search (title, description).
Multi-fields — title.raw keyword sub-field for exact sort while title remains searchable.

Dynamic mapping works for prototypes; VaultCommerce locks mapping before production indexing.

CRUD operations

Java

# Create (index)
PUT /vaultcommerce-products/_doc/SKU-8842
{"sku":"SKU-8842","title":"Trail Pack 40L","price":89.0,"category":"outdoor"}

# Read
GET /vaultcommerce-products/_doc/SKU-8842

# Partial update
POST /vaultcommerce-products/_update/SKU-8842
{"doc":{"price":79.0,"in_stock":true}}

# Delete
DELETE /vaultcommerce-products/_doc/SKU-8842

Java

from elasticsearch import Elasticsearch

es = Elasticsearch("http://localhost:9200")

doc = {"sku": "SKU-8842", "title": "Trail Pack 40L", "price": 89.0}
es.index(index="vaultcommerce-products", id="SKU-8842", document=doc)
hit = es.get(index="vaultcommerce-products", id="SKU-8842")
es.update(index="vaultcommerce-products", id="SKU-8842", doc={"price": 79.0})

Bulk import for catalog sync

Java

POST /_bulk
{ "index": { "_index": "vaultcommerce-products", "_id": "SKU-8842" } }
{ "sku": "SKU-8842", "title": "Trail Pack 40L", "price": 89.0 }
{ "index": { "_index": "vaultcommerce-products", "_id": "SKU-3310" } }
{ "sku": "SKU-3310", "title": "Summit Tent", "price": 249.0 }

Use bulk batches of 1,000–5,000 documents with refresh=false during large imports, then force refresh once.

Quick recall

Everything you need if you only revisit this box.

Mapping defines field types — keyword for exact match, text for analyzed search.
CRUD: index (PUT/POST), get, update (_update), delete — documents are JSON.
VaultCommerce maps SKU/category as keyword, title/description as text.
Bulk _bulk API is mandatory for large catalog imports.
Explicit mapping beats dynamic guessing — mistakes require reindex.
Multi-fields (title.raw) support both search and exact sort.

Test yourself

Answer these before moving on — recall is what makes it stick.