Driver 01
Licensing and hosting choice
Elasticsearch is offered under SSPL, the Elastic License and, since 2024, AGPL; OpenSearch is Apache 2.0, which some legal teams and hosting options prefer.
Migration playbook: Elasticsearch → OpenSearch / ClickHouse
Elasticsearch to OpenSearch and ClickHouse migration
Separate search from log analytics: OpenSearch for full-text relevance and ClickHouse for high-volume logs, each sized for its own workload.
Migration drivers
Why teams make this move
Driver 02
Log storage cost
Inverted indexes and JVM memory make long log retention expensive.
Driver 03
Cluster operations
Shard rebalancing and garbage-collection pauses under heavy ingestion.
Execution sequence
How the migration runs
Each phase ends with a check you can verify — data parity, error rates, latency — and the rollback path is agreed before any traffic moves.
- 01Phase
Workload separation
Splitting data into full-text search (OpenSearch) and time-series analytics (ClickHouse).
- 02Phase
OpenSearch for search
Moving search indices with snapshot and restore, or a reindex, depending on the version.
- 03Phase
ClickHouse for logs
Streaming log events into MergeTree tables in batches.
- 04Phase
Query routing and shutdown
Pointing clients at the new stores and decommissioning the Elasticsearch nodes.
Risk prevention
Pitfalls that derail this migration
Risk 01
ClickHouse as a search engine
ClickHouse is excellent at filtering and aggregation, but OpenSearch remains better for fuzzy BM25 relevance.
Risk 02
Snapshot version gaps
Snapshot restores failing across major versions without an intermediate reindex.
Risk 03
Tiny inserts
Sending single-row inserts to ClickHouse instead of batches, which causes too-many-parts errors.
Before and after
What we measure
We take a baseline before any change and report the same numbers after cutover, from your own tools. They are the evidence of whether the migration worked — not figures promised in advance.
- Storage
- Disk per day of logs, Elasticsearch vs ClickHouse
- Query latency
- Top dashboard and search queries, before and after
- Monthly cost
- Cluster compute and storage, before and after
Questions
Frequently asked migration questions
Close, for 7.10-era features: OpenSearch was forked from Elasticsearch 7.10.2, so most queries and index settings carry over. Newer Elasticsearch client libraries refuse to connect to non-Elastic servers, so you switch to the OpenSearch clients, and features added to Elasticsearch after the fork need checking case by case.
Columnar storage compresses log data far better than an inverted index, and ClickHouse aggregates long time ranges quickly. You give up fuzzy full-text relevance, which is why search stays on OpenSearch.
Rehearse the cutover before the real one
Tell us about your data volume, traffic and timeline. An engineer will reply within one business day to set up a call about the migration plan and its rollback path.