# \#vector-search **URL:** https://discuss.elastic.co/tag/vector-search/116.md [Latest](https://discuss.elastic.co/latest.md) · [Categories](https://discuss.elastic.co/categories.md) · [Tags](https://discuss.elastic.co/tags.md) --- ## [Elasticsearch 8.17 → 8.19 upgrade: kNN now eagerly defaults "k = size", breaking aggregation-only queries](https://discuss.elastic.co/t/elasticsearch-8-17-8-19-upgrade-knn-now-eagerly-defaults-k-size-breaking-aggregation-only-queries/384628)
**Author:** [@shekhar\_k](https://discuss.elastic.co/u/shekhar_k)\ **Replies:** 5\ **Last updated:** [June 26, 2026, 7:11pm UTC](https://discuss.elastic.co/t/elasticsearch-8-17-8-19-upgrade-knn-now-eagerly-defaults-k-size-breaking-aggregation-only-queries/384628 "2026-06-26T19:11:15Z")
I’m upgrading Elasticsearch from 8.17.3 to 8.19.10 and ran into a behavioural change with kNN + aggregations that breaks an existing use case. What worked in 8.17.3: We use knn inside the query DSL (bool.must) together… --- ## [Eland-imported naver/splade-v3 text\_expansion produces much smaller sparse vectors and worse ranking than local SentenceTransformers SparseEncoder](https://discuss.elastic.co/t/eland-imported-naver-splade-v3-text-expansion-produces-much-smaller-sparse-vectors-and-worse-ranking-than-local-sentencetransformers-sparseencoder/386819)
**Author:** [@alrolo3](https://discuss.elastic.co/u/alrolo3)\ **Replies:** 0\ **Last updated:** [June 11, 2026, 9:39pm UTC](https://discuss.elastic.co/t/eland-imported-naver-splade-v3-text-expansion-produces-much-smaller-sparse-vectors-and-worse-ranking-than-local-sentencetransformers-sparseencoder/386819 "2026-06-11T21:39:41Z")
Eland-imported naver/splade-v3 text\_expansion produces much smaller sparse vectors and worse ranking than local SentenceTransformers SparseEncoder Environment Elasticsearch version: 9.4.2 Eland Docker image used: docke… --- ## [Elasticsearch Serverless + Vector Search + Data Streams / Time Series Tradeoffs](https://discuss.elastic.co/t/elasticsearch-serverless-vector-search-data-streams-time-series-tradeoffs/386275)
**Author:** [@Januka\_Samaranayake](https://discuss.elastic.co/u/Januka_Samaranayake)\ **Replies:** 3\ **Last updated:** [May 12, 2026, 8:16am UTC](https://discuss.elastic.co/t/elasticsearch-serverless-vector-search-data-streams-time-series-tradeoffs/386275 "2026-05-12T08:16:25Z")
Hello everyone, I am trying to better understand the tradeoffs between Elasticsearch Serverless, Data Streams, TSDS (index.mode=time\_series), and future vector search workloads. Current Situation We currently have arou… --- ## [Unique vs Multiple dense vectors](https://discuss.elastic.co/t/unique-vs-multiple-dense-vectors/385710)
**Author:** [@cilasmarques](https://discuss.elastic.co/u/cilasmarques)\ **Replies:** 1\ **Last updated:** [April 1, 2026, 12:58am UTC](https://discuss.elastic.co/t/unique-vs-multiple-dense-vectors/385710 "2026-04-01T00:58:30Z")
Hi there! I'm adding semantic search to my site's search functionality. I use Elasticsearch and my index is structured as follows: "title":{ "type":"text", "analyzer":"general\_analyzer", "fields": { "spell"… --- ## [Best way to store document chunks for vector search as production standard](https://discuss.elastic.co/t/best-way-to-store-document-chunks-for-vector-search-as-production-standard/385414)
**Author:** [@grunggy](https://discuss.elastic.co/u/grunggy)\ **Replies:** 2\ **Last updated:** [March 12, 2026, 5:43pm UTC](https://discuss.elastic.co/t/best-way-to-store-document-chunks-for-vector-search-as-production-standard/385414 "2026-03-12T17:43:33Z")
Hi, working on a RAG setup and trying to land on a sensible production architecture for chunk storage and retrieval. Curious what others are running at scale. Large documents get split into chunks at ingestion, each chu… --- ## [Dynamic template mapping overridden by automatic \`dense\_vector\` inference for float arrays](https://discuss.elastic.co/t/dynamic-template-mapping-overridden-by-automatic-dense-vector-inference-for-float-arrays/385359)
**Author:** [@giga811](https://discuss.elastic.co/u/giga811)\ **Replies:** 3\ **Last updated:** [March 7, 2026, 6:16am UTC](https://discuss.elastic.co/t/dynamic-template-mapping-overridden-by-automatic-dense-vector-inference-for-float-arrays/385359 "2026-03-07T06:16:54Z")
I encountered a situation where a dynamic template specifying a field as float is overridden by automatic dense\_vector mapping when indexing a long float array in Elasticsearch. My intention is to store embeddings as a … --- ## [How to Increase the throughput for KNN in my ES cluster](https://discuss.elastic.co/t/how-to-increase-the-throughput-for-knn-in-my-es-cluster/384252)
**Author:** [@Esteban\_Velasquez](https://discuss.elastic.co/u/Esteban_Velasquez)\ **Replies:** 6\ **Last updated:** [January 31, 2026, 10:44am UTC](https://discuss.elastic.co/t/how-to-increase-the-throughput-for-knn-in-my-es-cluster/384252 "2026-01-31T10:44:11Z")
Hello! I created a track in ESrally with some queries that are normally done in my ES cluster and I replicated the big index (using snapshot) from the original cluster I have into a new one. I am doing an ESrally test f… --- ## [Explanation about DiskBBQ parameters](https://discuss.elastic.co/t/explanation-about-diskbbq-parameters/384248)
**Author:** [@viperv](https://discuss.elastic.co/u/viperv)\ **Replies:** 2\ **Last updated:** [December 30, 2025, 5:09pm UTC](https://discuss.elastic.co/t/explanation-about-diskbbq-parameters/384248 "2025-12-30T17:09:43Z")
Hi, I am looking into using DiskBBQ. I’ve read this blog post, and I noticed there are multiple parameters that can be defined, but there isn’t a thorough explanation about them: cluster size - with a default value of … --- ## [How can I delay HNSW index construction until a segment reaches a target size?](https://discuss.elastic.co/t/how-can-i-delay-hnsw-index-construction-until-a-segment-reaches-a-target-size/383826)
**Author:** [@sfmqrb](https://discuss.elastic.co/u/sfmqrb)\ **Replies:** 1\ **Last updated:** [December 3, 2025, 7:38pm UTC](https://discuss.elastic.co/t/how-can-i-delay-hnsw-index-construction-until-a-segment-reaches-a-target-size/383826 "2025-12-03T19:38:50Z")
I’m trying to tune vector indexing behavior. Specifically, I want the first HNSW index for a segment to be built only after the segment reaches a certain size (e.g., X MB or Y vectors), instead of building small HNSW gra… --- ## [Understanding how Elasticsearch segments, merges, and compacts its vector indexes](https://discuss.elastic.co/t/understanding-how-elasticsearch-segments-merges-and-compacts-its-vector-indexes/383796)
**Author:** [@sfmqrb](https://discuss.elastic.co/u/sfmqrb)\ **Replies:** 0\ **Last updated:** [December 2, 2025, 12:30am UTC](https://discuss.elastic.co/t/understanding-how-elasticsearch-segments-merges-and-compacts-its-vector-indexes/383796 "2025-12-02T00:30:01Z")
In Elasticsearch’s vector indexing pipeline, how exactly do segmenting and segment compaction work? Specifically, what compaction policy does Elasticsearch follow for segments and their vector indexes, and during compact… --- ## [Dense\_vector slow in ES 9.1.3](https://discuss.elastic.co/t/dense-vector-slow-in-es-9-1-3/381855)
**Author:** [@glund0](https://discuss.elastic.co/u/glund0)\ **Replies:** 16\ **Last updated:** [October 13, 2025, 10:42am UTC](https://discuss.elastic.co/t/dense-vector-slow-in-es-9-1-3/381855 "2025-10-13T10:42:49Z")
We’ve been using ES 8.9.0 and are looking to upgrade, so I started testing performance with 9.1.3 and found it to be very slow. So, I went back to 8.19.3 and found that it’s performance is fine. But we’d like to not be… --- ## [Dense Vector Field Extremely Large](https://discuss.elastic.co/t/dense-vector-field-extremely-large/382380)
**Author:** [@nicky](https://discuss.elastic.co/u/nicky)\ **Replies:** 12\ **Last updated:** [October 6, 2025, 11:57pm UTC](https://discuss.elastic.co/t/dense-vector-field-extremely-large/382380 "2025-10-06T23:57:16Z")
Hi all, Have been experimenting with applying both forms of compression to our dense vectors and doing performance comparisons, but while bbq\_hnsw has been performing relatively well at 1-3s per query average, int8 has … --- ## [Dynamic Template usage for nested fields](https://discuss.elastic.co/t/dynamic-template-usage-for-nested-fields/382229)
**Author:** [@Shweta\_Chadha](https://discuss.elastic.co/u/Shweta_Chadha)\ **Replies:** 3\ **Last updated:** [September 26, 2025, 8:21pm UTC](https://discuss.elastic.co/t/dynamic-template-usage-for-nested-fields/382229 "2025-09-26T20:21:48Z")
Hi folks! We are on Elastic version 8.17 and have an existing ES index that we are now introducing a dynamic attribute using dynamic template to. The dynamic fields would be added within an existing nested field. Our br… --- ## [400-Bad request](https://discuss.elastic.co/t/400-bad-request/381344)
**Author:** [@tarun-ghcp](https://discuss.elastic.co/u/tarun-ghcp)\ **Replies:** 2\ **Last updated:** [August 27, 2025, 9:35am UTC](https://discuss.elastic.co/t/400-bad-request/381344 "2025-08-27T09:35:03Z")
Below is my payload but it works only when query\_body is empty. I couldn’t figure out whats wrong inside it. I have a Docker container for the MCP server and it throws Bad Request every time. payload = { "jsonrpc": "2.… --- ## [400 Bad request](https://discuss.elastic.co/t/400-bad-request/381357)
**Author:** [@tarun-ghcp](https://discuss.elastic.co/u/tarun-ghcp)\ **Replies:** 1\ **Last updated:** [August 27, 2025, 3:31am UTC](https://discuss.elastic.co/t/400-bad-request/381357 "2025-08-27T03:31:14Z")
I am using Docker container for MCP server. When running my Python script to do a knn search, the server throws 400 Bad request. But when request is sent without anything inside query\_body, it gives me the result. I coul… --- ## [The RRF retriever could use a weighting option](https://discuss.elastic.co/t/the-rrf-retriever-could-use-a-weighting-option/378182)
**Author:** [@jan.stap](https://discuss.elastic.co/u/jan.stap)\ **Replies:** 2\ **Last updated:** [August 12, 2025, 2:19pm UTC](https://discuss.elastic.co/t/the-rrf-retriever-could-use-a-weighting-option/378182 "2025-08-12T14:19:06Z")
Hi! I would like to combine a multi\_match query with two kNN queries. For that I use the RRF retriever, which itself uses a Standard retriever and two kNN retrievers. This works fine, but I have no option to add weighti… --- ## [Snapshot Recovery Results in Empty Index for Dense Vector Embeddings (v8.18.2)](https://discuss.elastic.co/t/snapshot-recovery-results-in-empty-index-for-dense-vector-embeddings-v8-18-2/380168)
**Author:** [@Khanh\_Chuong\_Le](https://discuss.elastic.co/u/Khanh_Chuong_Le)\ **Replies:** 1\ **Last updated:** [August 12, 2025, 1:33pm UTC](https://discuss.elastic.co/t/snapshot-recovery-results-in-empty-index-for-dense-vector-embeddings-v8-18-2/380168 "2025-08-12T13:33:02Z")
I am experiencing an issue with Elasticsearch version 8.18.2 on a self-managed setup. I've set up an index with vector embeddings, templated as dense vectors. The index contains thousands of documents and search operatio… --- ## [Using dense\_vector with script params](https://discuss.elastic.co/t/using-dense-vector-with-script-params/380908)
**Author:** [@joellupfer](https://discuss.elastic.co/u/joellupfer)\ **Replies:** 1\ **Last updated:** [August 8, 2025, 3:07pm UTC](https://discuss.elastic.co/t/using-dense-vector-with-script-params/380908 "2025-08-08T15:07:46Z")
Hello! I am currently developing a script that utilizes text embedding vectors. The relevant portion of the mapping is shown below (please disregard the remaining parts, as this is intended solely for testing purposes). … --- ## [Would int8\_hnsw slower than hnsw for vector search](https://discuss.elastic.co/t/would-int8-hnsw-slower-than-hnsw-for-vector-search/380290)
**Author:** [@wangbing993](https://discuss.elastic.co/u/wangbing993)\ **Replies:** 5\ **Last updated:** [July 23, 2025, 2:57pm UTC](https://discuss.elastic.co/t/would-int8-hnsw-slower-than-hnsw-for-vector-search/380290 "2025-07-23T14:57:53Z")
Data result by VectorDBBench: Elastic Search Version: 8.14.3 mappings config: {'\_source': {'excludes': \['vector'\]}, 'properties': {'id': {'type': 'integer', 'store': True}, 'vector': {'dims': 1024, 'type': 'dense\_ve… --- ## [Vector search large dense vectors performance issues](https://discuss.elastic.co/t/vector-search-large-dense-vectors-performance-issues/380141)
**Author:** [@Pablo\_Delgado](https://discuss.elastic.co/u/Pablo_Delgado)\ **Replies:** 3\ **Last updated:** [July 16, 2025, 1:42pm UTC](https://discuss.elastic.co/t/vector-search-large-dense-vectors-performance-issues/380141 "2025-07-16T13:42:45Z")
I experience inconsistent (slow and fast results) on vector search. I currently have 50 million documents in an index, the vectors are stored in a filed with the current mapping: "large\_1536\_embedding": { "type": "de… --- ## [ No Observable Difference Between BBQ and Default Configurations in Elasticsearch – Help with Index Size Comparison](https://discuss.elastic.co/t/no-observable-difference-between-bbq-and-default-configurations-in-elasticsearch-help-with-index-size-comparison/377817)
**Author:** [@mohab\_ghobashy](https://discuss.elastic.co/u/mohab_ghobashy)\ **Replies:** 15\ **Last updated:** [July 2, 2025, 3:03pm UTC](https://discuss.elastic.co/t/no-observable-difference-between-bbq-and-default-configurations-in-elasticsearch-help-with-index-size-comparison/377817 "2025-07-02T15:03:15Z")
I've been running some tests on Better Binary Quantization (BBQ) in Elasticsearch and comparing it with the default configuration for dense vectors, but I'm not observing the expected differences in disk size or search p… --- ## [Top-Level knn with collapse: Accuracy-Performance Trade-off & Filter Behavior](https://discuss.elastic.co/t/top-level-knn-with-collapse-accuracy-performance-trade-off-filter-behavior/379349)
**Author:** [@Rami\_Salman](https://discuss.elastic.co/u/Rami_Salman)\ **Replies:** 1\ **Last updated:** [June 20, 2025, 7:36am UTC](https://discuss.elastic.co/t/top-level-knn-with-collapse-accuracy-performance-trade-off-filter-behavior/379349 "2025-06-20T07:36:50Z")
Hello Elasticsearch Community, I'm facing a critical challenge trying to balance search accuracy and performance when combining kNN queries with collapse and inner\_hits in Elasticsearch 8.17. I'm seeing a puzzling filte… --- ## [Not able to get more than top 4 result matches, regardless of configuration](https://discuss.elastic.co/t/not-able-to-get-more-than-top-4-result-matches-regardless-of-configuration/378546)
**Author:** [@dave\_espinosa\_qs](https://discuss.elastic.co/u/dave_espinosa_qs)\ **Replies:** 3\ **Last updated:** [May 28, 2025, 10:29am UTC](https://discuss.elastic.co/t/not-able-to-get-more-than-top-4-result-matches-regardless-of-configuration/378546 "2025-05-28T10:29:12Z")
Hello everyone, I am using ES Enterprise version 8.18.1 (I cannot change it for the time being BTW, as that is managed by other department). I created a Vector Database - like ES Vector Store, with the following Mapping… --- ## [Creating vector index from SharePoint Online data](https://discuss.elastic.co/t/creating-vector-index-from-sharepoint-online-data/378000)
**Author:** [@UNNIE\_Ayilliath](https://discuss.elastic.co/u/UNNIE_Ayilliath)\ **Replies:** 0\ **Last updated:** [May 9, 2025, 9:35pm UTC](https://discuss.elastic.co/t/creating-vector-index-from-sharepoint-online-data/378000 "2025-05-09T21:35:46Z")
I am trying to implement a vector index using ELSER2. The data source is SharePoint Online and I have used the Elasticsearch SharePoint Online connector to ingest data into the index with DLS enabled. I followed this bl… --- ## [Adaptive Replica Selection and knn load balancing](https://discuss.elastic.co/t/adaptive-replica-selection-and-knn-load-balancing/377991)
**Author:** [@peedeeboy](https://discuss.elastic.co/u/peedeeboy)\ **Replies:** 0\ **Last updated:** [May 9, 2025, 1:55pm UTC](https://discuss.elastic.co/t/adaptive-replica-selection-and-knn-load-balancing/377991 "2025-05-09T13:55:54Z")
hey friends :wave: I posted previously about our efforts on optimising our dedicated knn cluster. The latest thing we've been trying to understand / solve is why when we run load/stress testing, we often see 1 or 2 nod… --- ## [Performance Issue with KNN + Filter on Large Index (v8.12)](https://discuss.elastic.co/t/performance-issue-with-knn-filter-on-large-index-v8-12/377279)
**Author:** [@Saleh\_AbuAli](https://discuss.elastic.co/u/Saleh_AbuAli)\ **Replies:** 12\ **Last updated:** [April 28, 2025, 12:48pm UTC](https://discuss.elastic.co/t/performance-issue-with-knn-filter-on-large-index-v8-12/377279 "2025-04-28T12:48:33Z")
Hi, I’m currently using KNN search in Elasticsearch (version 8.12) with the following setup: Query vector dimension: 384 Index size: 200M+ documents k = 100, num\_candidates = 200 Quantized vectors using 'int8\_hnsw' in… --- ## [Duplicate indexing behavior without \_id](https://discuss.elastic.co/t/duplicate-indexing-behavior-without-id/377501)
**Author:** [@csanadpoda](https://discuss.elastic.co/u/csanadpoda)\ **Replies:** 1\ **Last updated:** [April 24, 2025, 9:07pm UTC](https://discuss.elastic.co/t/duplicate-indexing-behavior-without-id/377501 "2025-04-24T21:07:06Z")
If I were to index documents without specifying a fixed \_id, when duplicate documents appear, would they be created as duplicate entries in the index, or would ES recognize that there's already the same entry in the inde… --- ## [Implementing Relative Score Fusion for Hybrid Search in Elasticsearch](https://discuss.elastic.co/t/implementing-relative-score-fusion-for-hybrid-search-in-elasticsearch/364408)
**Author:** [@ayubSubhaniya](https://discuss.elastic.co/u/ayubSubhaniya)\ **Replies:** 3\ **Last updated:** [April 24, 2025, 7:50pm UTC](https://discuss.elastic.co/t/implementing-relative-score-fusion-for-hybrid-search-in-elasticsearch/364408 "2025-04-24T19:50:05Z")
Hello Elasticsearch Community, I'm interested in implementing Relative Score Fusion (RSF) directly in Elasticsearch to combine BM25 and kNN search results with weighted rankings based on query length or type. I want to … --- ## [When Does BBQ Quantization Outperform Scalar Quantization](https://discuss.elastic.co/t/when-does-bbq-quantization-outperform-scalar-quantization/377393)
**Author:** [@DylanWelzel](https://discuss.elastic.co/u/DylanWelzel)\ **Replies:** 1\ **Last updated:** [April 23, 2025, 12:32pm UTC](https://discuss.elastic.co/t/when-does-bbq-quantization-outperform-scalar-quantization/377393 "2025-04-23T12:32:26Z")
Hi all, I’m experimenting with the new vector quantization formats in Elasticsearch 8.x and trying to figure out at what dataset size BBQ (binary quantization) really starts to outperform scalar quantization. What I’m … --- ## [Trouble implementing Elasticsearch BBQ](https://discuss.elastic.co/t/trouble-implementing-elasticsearch-bbq/377269)
**Author:** [@DylanWelzel](https://discuss.elastic.co/u/DylanWelzel)\ **Replies:** 1\ **Last updated:** [April 18, 2025, 5:46am UTC](https://discuss.elastic.co/t/trouble-implementing-elasticsearch-bbq/377269 "2025-04-18T05:46:02Z")
I am trying to integrate BBQ, specifically bbq\_hnsw into my existing index. However I'm not seeing the performance benefits I would expect. { "type": "dense\_vector", "dims": dims, "index": True, … [Next page](https://discuss.elastic.co/tag/vector-search/116.md?match_all_tags=true&page=1&tags%5B%5D=vector-search)