--- title: Get to know OpenSearch 3.9 | daily.dev description: OpenSearch 3.9 introduces a native C++ engine for neural sparse vector search that boosts throughput by 39% and shrinks JVM heap usage 8x, along with new... canonical: https://daily.dev/posts/get-to-know-opensearch-3-9-qjzsnythj twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Get to know OpenSearch 3.9 | daily.dev og:description: OpenSearch 3.9 introduces a native C++ engine for neural sparse vector search that boosts throughput by 39% and shrinks JVM heap usage 8x, along with new... og:url: https://daily.dev/posts/get-to-know-opensearch-3-9-qjzsnythj og:image: https://api.daily.dev/og/posts/qjZsNyTHj.png og:image:alt: Get to know OpenSearch 3.9 og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Get to know OpenSearch 3.9 **[OpenSearch](https://daily.dev/sources/opensearch)** · 17 min read · 0 upvotes · 0 comments ## Summary OpenSearch 3.9 introduces a native C++ engine for neural sparse vector search that boosts throughput by 39% and shrinks JVM heap usage 8x, along with new half_float and bf16 vector compression options, plus 2-bit/4-bit scalar quantization. The unified alerts view now includes anomaly detection and forecasting and is generally available. The separately versioned OpenSearch Observability Stack gets redesigned trace details, reusable PromQL dashboards, no-code alert rule building, a guided APM setup wizard, and PPL query profiling and linting. On resiliency, adaptive per-action concurrency limits and index-level search pruning (cutting tail latency over 80% in benchmarks) help clusters handle load, and the resource sharing and access control framework reaches general availability. Amazon Linux 2 support is deprecated as a build image and OS target. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What performance improvement does the new native engine in OpenSearch 3.9 provide for neural sparse ANN search? OpenSearch 3.9's native C++ engine for neural sparse approximate nearest neighbor search delivered 39% higher throughput, 3.3x faster index builds, and an 8x smaller JVM heap footprint in testing on an 8.8-million-document corpus. It runs the SEISMIC algorithm via the Java Native Interface, memory-mapping index data instead of holding it on the Java heap. Enable it by setting "engine": "native" in the sparse_vector field mapping; the Lucene engine remains the default. _Teams tuning vector search performance can track engine-level changes like this one on daily.dev._ ### What is the difference between the half_float and bf16 vector storage options added in OpenSearch 3.9? Both are 16-bit formats that halve memory versus 32-bit float, but half_float uses standard FP16 while bf16 keeps the same eight exponent bits as 32-bit float, giving it a wider value range without clipping at the cost of slightly lower recall on some datasets. bf16 is accelerated by AVX-512 BF16 instructions on Intel Sapphire Rapids and newer CPUs. Both work in Faiss and Lucene engines with HNSW and flat index structures. _Engineers weighing vector compression trade-offs can follow storage format updates like these on daily.dev._ ### Is Amazon Linux 2 still supported by OpenSearch? OpenSearch 3.9.0 deprecates support for Amazon Linux 2 as both a CI build image and a supported operating system, since Amazon Linux 2 itself reached end of support on June 30, 2026. Teams still running OpenSearch on Amazon Linux 2 should check the supported operating systems documentation and plan a migration to a compatible OS. _Ops teams planning OS migrations can keep tabs on deprecations like this one via daily.dev._ ## Similar posts on daily.dev - [Amazon OpenSearch Service now supports OpenSearch version 3.7](https://daily.dev/posts/amazon-opensearch-service-now-supports-opensearch-version-3-7-wkqtg7wcc) · AWS · 0 upvotes · 0 comments - [Inside OpenSearch’s bid to become the default AI data layer](https://daily.dev/posts/inside-opensearch-s-bid-to-become-the-default-ai-data-layer-2uzzh37ue) · The New Stack · 2 upvotes · 0 comments - [OpenSearch vs Elasticsearch Compared: Performance, Features & Cost \(2026\)](https://daily.dev/posts/opensearch-vs-elasticsearch-compared-performance-features-cost-2026--vufwqexwq) · BigData Boutique blog · 1 upvotes · 0 comments - [Uber Adopts Amazon OpenSearch for Semantic Search to Better Capture User Intent](https://daily.dev/posts/uber-adopts-amazon-opensearch-for-semantic-search-to-better-capture-user-intent-2caemtik8) · InfoQ · 1 upvotes · 0 comments --- Tags: [#data-science](https://daily.dev/tags/data-science), [#observability](https://daily.dev/tags/observability), [#vector-search](https://daily.dev/tags/vector-search), [#opentelemetry](https://daily.dev/tags/opentelemetry), [#opensearch](https://daily.dev/tags/opensearch) [View this post on daily.dev](https://daily.dev/posts/get-to-know-opensearch-3-9-qjzsnythj) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Get to know OpenSearch 3.9","url":"https://daily.dev/posts/get-to-know-opensearch-3-9-qjzsnythj","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/get-to-know-opensearch-3-9-qjzsnythj"},"datePublished":"2026-09-29T23:30:12.073Z","dateModified":"2026-09-29T23:30:38.658Z","description":"OpenSearch 3.9 introduces a native C++ engine for neural sparse vector search that boosts throughput by 39% and shrinks JVM heap usage 8x, along with new...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/9b879f497b0c52adaa19d267963738a3?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/9b879f497b0c52adaa19d267963738a3?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"OpenSearch","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"OpenSearch","logo":"https://media.daily.dev/image/upload/s--QsmPuJ8K--/f_auto,q_auto/v1787487638/logos/opensearch-blog?_a=BAMAMicg0","url":"https://daily.dev/sources/opensearch"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/get-to-know-opensearch-3-9-qjzsnythj","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"data-science,observability,vector-search,opentelemetry,opensearch","timeRequired":"PT17M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"OpenSearch","item":"https://daily.dev/sources/opensearch"},{"@type":"ListItem","position":3,"name":"Get to know OpenSearch 3.9"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/get-to-know-opensearch-3-9-qjzsnythj#faq","mainEntity":[{"@type":"Question","name":"What performance improvement does the new native engine in OpenSearch 3.9 provide for neural sparse ANN search?","acceptedAnswer":{"@type":"Answer","text":"OpenSearch 3.9's native C++ engine for neural sparse approximate nearest neighbor search delivered 39% higher throughput, 3.3x faster index builds, and an 8x smaller JVM heap footprint in testing on an 8.8-million-document corpus. It runs the SEISMIC algorithm via the Java Native Interface, memory-mapping index data instead of holding it on the Java heap. Enable it by setting \"engine\": \"native\" in the sparse_vector field mapping; the Lucene engine remains the default. Teams tuning vector search performance can track engine-level changes like this one on daily.dev."}},{"@type":"Question","name":"What is the difference between the half_float and bf16 vector storage options added in OpenSearch 3.9?","acceptedAnswer":{"@type":"Answer","text":"Both are 16-bit formats that halve memory versus 32-bit float, but half_float uses standard FP16 while bf16 keeps the same eight exponent bits as 32-bit float, giving it a wider value range without clipping at the cost of slightly lower recall on some datasets. bf16 is accelerated by AVX-512 BF16 instructions on Intel Sapphire Rapids and newer CPUs. Both work in Faiss and Lucene engines with HNSW and flat index structures. Engineers weighing vector compression trade-offs can follow storage format updates like these on daily.dev."}},{"@type":"Question","name":"Is Amazon Linux 2 still supported by OpenSearch?","acceptedAnswer":{"@type":"Answer","text":"OpenSearch 3.9.0 deprecates support for Amazon Linux 2 as both a CI build image and a supported operating system, since Amazon Linux 2 itself reached end of support on June 30, 2026. Teams still running OpenSearch on Amazon Linux 2 should check the supported operating systems documentation and plan a migration to a compatible OS. Ops teams planning OS migrations can keep tabs on deprecations like this one via daily.dev."}}]} ```