OpenSearch for Confluence Data Center
This page explains the differences between OpenSearch and Lucene search platforms. By default, Confluence uses Lucene.
We recommend switching to OpenSearch because it offers performance, scalability, and reliability improvements over Lucene. You can use OpenSearch starting from Confluence 9.2.
As your Confluence site grows, you need a search that keeps up with your team’s pace. That’s why Confluence Data Center now offers OpenSearch as an opt-in alternative to the default Lucene search platform. OpenSearch is designed to scale with your organization, using multi-node instances to handle even the most process-intensive indexing. Your users will enjoy the same familiar search experience — just faster and more reliable, no matter how large your site becomes.
On this page:
- What is OpenSearch
- Differences between OpenSearch and Lucene
- Performance test results
- Ready to make the switch?
What is OpenSearch
OpenSearch is an open-source, distributed search and analytics suite. It began as a community-driven fork of Elasticsearch 7.10, created by Amazon Web Services (AWS) after Elastic changed its licensing model. OpenSearch offers advanced search features and real-time analytics.
With OpenSearch, you can:
Improve the performance of your instance. Discover the results of the performance tests
Simplify index management and lower overall Confluence resource consumption.
Keep search running without interruption on clustered OpenSearch instances, even during reindexing.
Quickly find the content you need with full-text search capabilities.
Gain real-time insights into your data and search patterns.
Customize and extend search functionality using apps and APIs.
Keep your data safe and stay informed with built-in security, alerting, and monitoring.
OpenSearch gives your Confluence instance operational benefits and improved search and analytics capabilities.
OpenSearch | Lucene | |
|---|---|---|
Operational differences | ||
Startup for new nodes | All nodes connect to a single, up-to-date instance, so new nodes start up quickly and are immediately ready to serve search requests. | When a new node joins a Lucene cluster, it must copy or rebuild the entire search index and replay recent changes. |
Search results | All changes are available to every node, reducing the risk of outdated or inconsistent search results. Updates are shared across the cluster after a short refresh interval (defaults to one second), minimizing the need for full site reindexes. | When a node in the cluster makes a change to a Lucene document, that change is firstly made to the local index, then propagated to the other nodes within the cluster. This means higher risk of outdated results and more overhead when scaling or adding nodes. |
Index scalability | All nodes connect to a dedicated OpenSearch cluster, so you can scale your search index independently from your Confluence cluster. This means you can use smaller, more cost-effective Confluence nodes, while scaling your search infrastructure as needed. | Every node must store a full copy of the index, requiring significant disk space and memory. |
Search and analytics differences | ||
Built-in search capabilities | Distributed architecture ensures that your search remains fast and reliable. You can extend its functionality with plugins and modules. | No built-in plugin or module system for extensibility. You can extend it by writing custom Java code. |
Trace analytics | You can monitor search indexes independently from their applications. Trace Analytics provides detailed analysis and visualization of distributed trace data, helping you identify bottlenecks and optimize performance. This is valuable for debugging and optimizing microservices architectures. | No built-in trace analytics or distributed trace visualization. |
Index management | OpenSearch automates index handling with policies for deletion. This enables smarter resource utilization and effective data lifecycle management, simplifying administrative tasks and keeping your search environment organized. | Index management is manual, you must handle deletion and cleanup yourself. No built-in monitoring or policy-driven automation for index lifecycle, making it more labor-intensive and error-prone as data grows. |
Performance test results
OpenSearch boosts overall Confluence performance and reliability. Our internal testing shows that OpenSearch can significantly improve scale-related performance issues, especially as your instance grows.
Our benchmarking results show that OpenSearch is approximately 4.5 times faster than a comparable Lucene-based search. This performance gap widens as your dataset grows.
Test details: We benchmarked a 36GB index on an internal 3-node Confluence instance, comparing Lucene (8 CPU, 32GB RAM) with OpenSearch (2 CPU, 16GB RAM).
CQL search performance
We used the GitHub DC app performance toolkit to simulate 20,000 actions per hour from 200 concurrent users. The results:
検索プラットフォーム | Median response time (lower is better) |
|---|---|
Lucene (ベースライン) | 2.34 秒 |
OpenSearch | 0.66 秒 |
Full reindexing performance
We manually triggered a full reindex and found that OpenSearch completed the process faster than Lucene.
Although OpenSearch reindexing might not be significantly faster, the updates are shared across the cluster after a short refresh interval (by default, one second). This approach reduces the need for full site reindexes compared to Lucene.
検索プラットフォーム | 時間 (短いほど良い) |
|---|---|
Lucene | 4 時間 59 分 |
OpenSearch | 4 時間 36 分 |
Ready to make the switch?
OpenSearch is designed to help your Confluence site grow, perform, and stay reliable — no matter how large your team or your data becomes. If you’re ready to take advantage of faster search, easier scaling, and a more resilient platform, consider enabling OpenSearch for your Confluence Data Center instance.