Amazon CloudSearch - Runbook & Reference¶
中文 · GitHub source Facts verified against official AWS documentation: 2026-08-19
Overview¶
Amazon CloudSearch is a fully managed search service for building search solutions over large collections of data such as web pages, documents, forum posts, and product information. You create a search domain, upload data, and query through an HTTP search endpoint. Note: Amazon CloudSearch is no longer available to new customers; existing customers can continue using the service.
Key concepts¶
- Search domain: a domain contains your searchable data and the search instances that serve requests; create separate domains for separate collections.
- Indexing: CloudSearch indexes structured data and plain text; the index is deployed to one or more search instances.
- Search features: full-text search with language-specific text processing, boolean, prefix and range searches, term boosting, faceting, highlighting, and autocomplete suggestions.
- Endpoints: a configuration endpoint (per Region) manages domains; each domain has a document endpoint (
doc-<domain>-<id>...) for uploads and a search endpoint (search-<domain>-<id>...) for queries; results are JSON or XML. - Scaling: add/remove search instances as data volume and traffic change.
Common operations (AWS CLI)¶
# Create and manage a domain
aws cloudsearch create-domain --domain-name products
aws cloudsearch describe-domains --domain-names products
# Upload documents (JSON/XML batches via the document endpoint)
aws cloudsearchdomain upload-documents --endpoint-url https://doc-products-xxxxxxxxx.us-east-1.cloudsearch.amazonaws.com \
--documents file://batch.json --content-type application/json
# Index and search
aws cloudsearch index-documents --domain-name products
aws cloudsearchdomain search --endpoint-url https://search-products-xxxxxxxxx.us-east-1.cloudsearch.amazonaws.com \
--query "laptop" --query-parser simple --size 10
# Delete a domain
aws cloudsearch delete-domain --domain-name products
Best practices¶
- Define a clear indexing schema (fields, facets, suggester) before uploading large batches.
- Batch document uploads and index once per batch to reduce indexing overhead.
- Size search instances to your data volume and query traffic; monitor latency and scale out proactively.
- Use facet and suggester options for rich result UX; secure the search endpoint (SigV4 or access policies) when it should not be public.
- Existing customers: plan migration to current search services if you are starting new projects, since new customers cannot onboard.
Troubleshooting¶
| Symptom | Checks and fixes |
|---|---|
| Documents not searchable | Confirm documents uploaded to the document endpoint and index-documents ran successfully. |
| Search returns no results | Check the query parser, field names, and that the index is in ACTIVE state. |
| Domain endpoint unknown | Get the endpoints from describe-domains; they include account/domain identifiers. |
| Slow search | Increase search instances or reduce result size; review facet/aggregation usage. |
| Access denied | Sign requests with SigV4 or configure access policies for anonymous query endpoints. |
Limits¶
Domains per account, search instances per domain, document sizes, and API rates have quotas; service onboarding is limited to existing customers. See the Amazon CloudSearch endpoints and quotas page for current values.