Use cases
Migrating existing data
Migration Assistant for Amazon OpenSearch Service offers various options for migrating existing data, including detailed guidance on running a reindex-from-snapshot applicable across all supported migration routes, such as from Elasticsearch 5.6, 6.8, 7.10.2, or 7.17 to OpenSearch 2.19 or OpenSearch 3.x on Amazon OpenSearch Service or Amazon OpenSearch Serverless.
Near real-time migration of HTTP traffic between clusters
The solution offers you the option to capture data destined for a source cluster and store this data for reuse. A user can replay this data to an Amazon OpenSearch Service domain or Amazon OpenSearch Serverless collection in near real-time to migrate as soon as possible, or replay at a later time.
Replay traffic to multiple targets
The solution allows you to capture traffic for replay through multiple instances or in sequential runs, facilitating the validation of diverse cluster workloads and configurations.
Precise simulation of your cluster workloads
The solution allows users to capture and replay traffic either simultaneously with multiple instances, or in separate sequential runs. This feature aids in validating different cluster workloads and configurations. By default, the Traffic Replayer preserves the original concurrency and request rate to accurately simulate production loads, ensuring a fair like-for-like comparison.
Migrating from Apache Solr
Migration Assistant supports migrating from Apache Solr 6.x–9.x to OpenSearch 3.x on Amazon OpenSearch Service or Amazon OpenSearch Serverless. This includes document backfill from Solr backups with automatic schema translation (Solr schema.xml field types to OpenSearch mappings) and auto-detection of SolrCloud and standalone Solr deployments. Live traffic capture and replay is not supported for Solr sources.
Migrating to Amazon OpenSearch Serverless
Migration Assistant migrates to Amazon OpenSearch Serverless collections from every supported source. The solution auto-detects the collection type (SEARCH, TIMESERIES, or VECTORSEARCH) and adjusts behavior accordingly: SEARCH collections preserve source document IDs, while TIMESERIES and VECTORSEARCH collections use server-generated IDs. When migrating to a VECTORSEARCH collection, knn_vector field mappings are automatically converted to Faiss HNSW for serverless compatibility, and model_id references are removed.
Validate target cluster results
The solution facilitates user comparisons of source and target traffic in terms of accuracy and performance. It captures metrics and logs for analysis, providing users with the necessary confidence to migrate their production traffic to a new target.
