Magento 2 search with OpenSearch / Elasticsearch

MySQL LIKE queries don't scale to a multi-million-SKU catalog. Magento offloads search to a dedicated engine — OpenSearch or Elasticsearch — and that's the backbone of storefront findability.

Why a search engine, not SQL

Catalog search needs partial matches, faceted filtering, ranking, and typo tolerance across huge indexes. A document engine indexes products as documents and answers queries in milliseconds, where SQL would scan and join EAV tables painfully.

The index and the pipeline

Magento maintains a search index as documents in OpenSearch. The flow:

1. Products are read via collections (the EAV model from earlier).
2. The catalogsearch indexer transforms them into search documents.
3. Documents are pushed to OpenSearch.
4. Storefront search queries OpenSearch directly, with facets for price, category, and attributes.

Keep the indexer scheduled (or queue-driven) so new and updated products become searchable promptly.

Running it locally

Govard provisions the search engine as part of the stack — OpenSearch 3.0 in the default Magento 2 configuration — so your local search behaves like production. While the environment is up, you can rebuild the index from govard shell:

bin/magento indexer:reindex catalogsearch_fulltext

Operational notes

• Monitor index size and shard count as the catalog grows.
• A stale index is the usual cause of "new product not showing in search" — reindex first, debug later.
• Facet performance depends on attribute storage; index the attributes you filter on.

What's next in this series

We've covered the read paths. The hardest operational task is the upgrade — and the profiling that proves it's worth it. Next: safe Magento 2 upgrades and the profiler.

Next in this series: Upgrading Magento 2 safely + profiling — /blog/magento2-upgrade-profiling