Link Search Menu Expand Document Documentation Menu

You're viewing the current version of Migration Assistant documentation (Kubernetes/EKS-based). For the classic ECS/CDK-based version, see the classic documentation.

Is Migration Assistant right for you?

Whether Migration Assistant is right for you depends on your migration path, downtime target, and how much platform work you want to own yourself.

Migration Assistant is designed for teams that want a workflow-driven migration platform rather than a single-use upgrade procedure. It is especially useful when:

  • You need to migrate across one or more major versions in a single step.
  • You want to validate the target before cutover.
  • You need a repeatable backfill process with retries and progress tracking.
  • You want a zero-downtime option through Capture and Replay.

Compared with traditional upgrade methods, Migration Assistant reduces the amount of manual coordination required between snapshot creation, metadata changes, backfill, validation, and cutover.

Migration concepts

If you’re new to Migration Assistant, you see a few terms repeatedly. The migration phases overview explains how they fit together. The shortest definitions are:

  • Backfill – Bulk migration of historical documents from a snapshot. Used for both planned-downtime and zero-downtime migrations.
  • Reindex-from-Snapshot (RFS) – The mechanism Migration Assistant uses for backfill. RFS reads shard data from a snapshot in object storage instead of querying the live source cluster, which is why it scales well and keeps load off the source.
  • Capture and Replay – Zero-downtime path that captures live writes from the source through a proxy, buffers them in Kafka, and replays them against the target after backfill catches up.
  • Migration phases – The ordered steps the workflow runs: assess, deploy, migrate metadata, backfill, optional Capture and Replay, validate, and switch traffic to the target.

Supported migration paths

The following matrix shows which source versions can be directly migrated to which OpenSearch target versions.

Source version OpenSearch 1.x OpenSearch 2.x OpenSearch 3.x
Elasticsearch 1.x–2.x Yes* Yes* Yes*
Elasticsearch 5.x–7.x Yes Yes Yes
Elasticsearch 8.x No Yes Yes
OpenSearch 1.x–2.x No Yes Yes
Apache Solr 6.x–9.x No No Yes*

* Backfill only—Capture and Replay is not supported for these source versions.

Version-specific notes

Elasticsearch 6.x: Elasticsearch 6.x generally uses single-type indexes, but upgraded or legacy datasets may still contain mappings that need type-handling decisions. Run metadata evaluation first; set multiTypeBehavior only if the evaluation reports multi-type mapping issues.

Elasticsearch 8.x: Supported with compatibility support for post-fork features. Some 8.x-specific features may not have OpenSearch equivalents. Test metadata migration first.

Apache Solr 6.x–9.x: Only the backfill migration approach is supported (no Capture and Replay). Migration Assistant auto-detects whether the source is SolrCloud or standalone. For more information, see Solr migration.

Supported platforms

The following table lists supported source and target platforms.

Platform Source Target
Self-managed (on-premises) Yes Yes
Amazon OpenSearch Service Yes Yes
Amazon OpenSearch Serverless No Yes
Third-party cloud providers Yes Yes
AWS EC2 Yes Yes
Apache Solr (SolrCloud/Standalone) Yes No

Deployment options

Migration Assistant runs on Kubernetes and can be deployed to:

  • Amazon EKS for the recommended AWS production path with bootstrap automation, pod identity, image mirroring, snapshot helpers, and CloudWatch integration.
  • Any Kubernetes cluster when you already operate your own Kubernetes platform or you are evaluating locally.

The migration engine is the same in both cases. The difference is how much of the surrounding platform is prepared for you.

Component support

The following table lists the components that Migration Assistant can migrate automatically and those that require manual migration.

Component Supported Recommendation
Documents Yes Migrate using RFS (backfill) or Capture and Replay
Index settings Yes Migrated automatically
Index mappings Yes Migrated automatically
Index templates Yes Migrated automatically
Component templates Yes Migrated automatically
Aliases Yes Migrated automatically
Data streams No Manually recreate on target
ISM/ILM policies No Manually recreate on target
Security configuration No Configure separately on target
Kibana/Dashboards objects No Export/import using Dashboards UI
Ingest pipelines No Manually recreate
Cluster settings No Configure separately

Checklist

Use this checklist to determine whether Migration Assistant is the right fit:

  • Are you migrating across one or more major versions in a single step?
  • Do you need to maintain high service availability with minimal or zero downtime?
  • Do you need to validate a new OpenSearch cluster before switching over?
  • Are you looking for tooling to migrate index settings and other metadata?
  • Do you need a high-performance backfill solution with pause, resume, and checkpoint recovery?
  • Are you migrating from Apache Solr and need a snapshot-based backfill solution?

Use Amazon EKS if you also want the deployment tooling to prepare the AWS environment around the migration.

If you answered “yes” to most of these questions, Migration Assistant is likely the right solution.

Assumptions and limitations

Migration Assistant has the following assumptions and limitations.

Reindex-from-Snapshot

For Elasticsearch and OpenSearch sources:

  • The source cluster must have the repository-s3 plugin installed (for S3-based snapshots).
  • Shards of up to 80 GiB are supported by default. This can be configured to support larger shards up to the limits of your EBS storage, except in AWS GovCloud regions which are limited to 80 GiB.
  • Snapshots of indexes using zstd or zstd_no_dict codecs (OpenSearch 2.9+) are not supported—reindex with default or best_compression first.

For Apache Solr sources:

  • The source cluster must have the Solr S3 backup plugin installed and a backup repository configured in solr.xml. Solr writes the backup directly to S3, and Migration Assistant reads it from there. See the Solr backfill guide for the full prerequisite list.

Capture and Replay

Capture and Replay has the following limitations:

  • Automatically generated document IDs are not preserved during replay—clients must explicitly provide document IDs to maintain consistency between source and target.
  • Live capture is recommended only for workloads with < 4 TB/day of incoming traffic.

Networking

The following networking requirements apply:

  • The Kubernetes cluster must have network connectivity to both source and target clusters.
  • For EKS deployments, source and target cluster security groups must allow inbound traffic from the EKS cluster security group.

Pre-migration checklist

Complete the following steps before starting a migration:

  • Verify source and target versions are in the preceding compatibility matrix.
  • Identify unsupported components and plan manual migration.
  • Plan index scope using index allow lists.
  • Test with a subset of 1–2 representative indexes first.
  • Verify whether multi-type indexes exist (ES 5.x and 6.x).