From 120af4475ab7284bb3f1d318785039f56fe7845a Mon Sep 17 00:00:00 2001 From: prrao87 <35005448+prrao87@users.noreply.github.com> Date: Fri, 13 Mar 2026 17:11:57 -0400 Subject: [PATCH 1/4] Update benchmarks/perf page for enterprise --- docs/docs.json | 2 +- docs/enterprise/benchmarks.mdx | 76 ------------------------------ docs/enterprise/performance.mdx | 82 +++++++++++++++++++++++++++++++++ 3 files changed, 83 insertions(+), 77 deletions(-) delete mode 100644 docs/enterprise/benchmarks.mdx create mode 100644 docs/enterprise/performance.mdx diff --git a/docs/docs.json b/docs/docs.json index 81a4b98..e23a4fe 100644 --- a/docs/docs.json +++ b/docs/docs.json @@ -50,7 +50,7 @@ "enterprise/quickstart", "enterprise/architecture", "enterprise/security", - "enterprise/benchmarks", + "enterprise/performance", { "group": "Deployment", "pages": [ diff --git a/docs/enterprise/benchmarks.mdx b/docs/enterprise/benchmarks.mdx deleted file mode 100644 index a293ea4..0000000 --- a/docs/enterprise/benchmarks.mdx +++ /dev/null @@ -1,76 +0,0 @@ ---- -title: "Benchmarks" -sidebarTitle: "Benchmarks" -description: "See numbers from LanceDB Enterprise's performance scalability and latency benchmarks." -icon: "tachometer-alt" ---- - -LanceDB's architecture is designed to deliver **25ms** vector search latency. -Even with metadata filtering, our query latency remains as low as **50ms**. -It is important to note that we can support thousands of QPS with such query performance. - -| Percentile | Vector Search | Vector Search w. Filtering | Full-Text Search | -|:------------:|:------------:|:-------------------------:|:----------------:| -| P50 | 25ms | 30ms | 26ms | -| P90 | 26ms | 39ms | 37ms | -| P99 | 35ms | 50ms | 42ms | - -## Dataset - -We used two datasets for this benchmark test: the [dbpedia-entities-openai-1M](https://huggingface.co/datasets/KShivendu/dbpedia-entities-openai-1M) -for vector search, and a synthetic dataset for vector search with metadata filtering. - -| Name | # Vectors | Vector Dimension | -|:------|:---------:|:---------------:| -| dbpedia-entities-openai-1M | 1,000,000 | 1536 | -| synthetic dataset | 15,000,000 | 256 | - -## Vector Search - -We ran vector queries with dbpedia-entities-openai-1M with a warmed-up cache. -The query latency is as follows: - -| Percentile | Latency | -|:------------:|:-------:| -| P50 | 25ms | -| P90 | 26ms | -| P99 | 35ms | -| Max | 49ms | - -## Full-Text Search - -With the same dataset and a warmed-up cache, the full-text search performance is as follows: - -| Percentile | Latency | -|:------------:|:-------:| -| P50 | 26ms | -| P90 | 37ms | -| P99 | 42ms | -| Max | 98ms | - -## Vector Search with Metadata Filtering - -We created a 15M-vector dataset with sufficient complexity to thoroughly test our complex metadata filtering capabilities. -Such filtering can span a wide range of scalar columns, e.g., "find Sci-fi movies since 1900". - -With a warmed-up cache, the query performance using slightly more selective filters, -e.g., "find Sci-fi movies between the years 2000 and 2012", is as follows: - -| Percentile | Latency | -|:-----------:|:-------:| -| P50 | 30ms | -| P90 | 39ms | -| P99 | 50ms | - -The query performance using complex filters, e.g., "find Sci-fi movies since 1900", is as follows: - -| Percentile | Latency | -|:------------:|:-------:| -| P50 | 65ms | -| P90 | 76ms | -| P99 | 100ms | - - - -Our benchmarks are designed to provide consistent and reproducible performance evaluations of LanceDB. We regularly update and re-run these benchmarks to ensure the data remains accurate and relevant. - diff --git a/docs/enterprise/performance.mdx b/docs/enterprise/performance.mdx new file mode 100644 index 0000000..9d37b9b --- /dev/null +++ b/docs/enterprise/performance.mdx @@ -0,0 +1,82 @@ +--- +title: "Performance Characteristics" +sidebarTitle: "Performance" +description: "Understand LanceDB Enterprise performance characteristics, including representative latency and scalability benchmarks." +icon: "tachometer-alt" +--- + +LanceDB Enterprise is designed for low-latency, high-throughput search, but observed performance depends on factors such as dataset shape, index configuration, cache warmth, filter selectivity, concurrency, and cluster sizing. The figures on this page should be read as representative benchmark results for a specific test setup, not as universal guarantees for every deployment. + +In our benchmark environment, warmed-cache vector search reached around **25ms** P50 latency, and metadata-filtered search reached around **50ms** P99 latency for the filter pattern shown below. + + +If you want performance guidance for your own workload, reach out to [contact@lancedb.com](mailto:contact@lancedb.com). The LanceDB engineering team can help map your latency, throughput, ingestion, and filtering requirements to an appropriate Enterprise cluster design. + + +| Percentile | Vector Search | Vector Search w. Filtering | Full-Text Search | +|:------------:|:------------:|:-------------------------:|:----------------:| +| P50 | 25ms | 30ms | 26ms | +| P90 | 26ms | 39ms | 37ms | +| P99 | 35ms | 50ms | 42ms | + +Depending on workload and tuning, Enterprise clusters can also be configured for high concurrency, including thousands of QPS in some deployments, but the right configuration varies by use case. Training, search, and analytics workloads often benefit from different cluster shapes and resource allocation strategies. To understand which parts of the system influence these results, see the [Enterprise architecture](/enterprise/architecture) guide. + +## Dataset + +We used two datasets for this benchmark: the [dbpedia-entities-openai-1M](https://huggingface.co/datasets/KShivendu/dbpedia-entities-openai-1M) +for vector search, and a synthetic dataset for vector search with metadata filtering. + +| Name | # Vectors | Vector Dimension | +|:------|:---------:|:---------------:| +| dbpedia-entities-openai-1M | 1,000,000 | 1536 | +| synthetic dataset | 15,000,000 | 256 | + +These benchmark results are most useful as a directional baseline. Different data distributions, index choices, cache behavior, and cluster settings can materially change the latency profile. + +## Vector Search + +We ran vector queries against `dbpedia-entities-openai-1M` with a warmed-up cache. In that benchmark setup, we observed the following latency profile: + +| Percentile | Latency | +|:------------:|:-------:| +| P50 | 25ms | +| P90 | 26ms | +| P99 | 35ms | +| Max | 49ms | + +## Full-Text Search + +With the same dataset and a warmed-up cache, full-text search fell into the following range: + +| Percentile | Latency | +|:------------:|:-------:| +| P50 | 26ms | +| P90 | 37ms | +| P99 | 42ms | +| Max | 98ms | + +## Vector Search with Metadata Filtering + +We created a 15M-vector dataset to evaluate metadata-aware search under more complex filtering conditions. +These filters can span a wide range of scalar columns, for example, "find Sci-fi movies since 1900". + +With a warmed-up cache, slightly more selective filters, for example, "find Sci-fi movies between the years 2000 and 2012", produced the following representative results: + +| Percentile | Latency | +|:-----------:|:-------:| +| P50 | 30ms | +| P90 | 39ms | +| P99 | 50ms | + +Broader or less selective filters, for example, "find Sci-fi movies since 1900", moved the latency range to: + +| Percentile | Latency | +|:------------:|:-------:| +| P50 | 65ms | +| P90 | 76ms | +| P99 | 100ms | + + + +These benchmarks are intended to provide consistent, reproducible reference points for LanceDB Enterprise rather than one-size-fits-all promises. We periodically re-run and update numbers as necessary, but production performance will still depend on workload shape, cluster tuning, and the architectural choices described in the [Enterprise architecture](/enterprise/architecture) guide. + From 42b5584afc8931427c68517dfe4e4edf8b4fb82a Mon Sep 17 00:00:00 2001 From: prrao87 <35005448+prrao87@users.noreply.github.com> Date: Fri, 13 Mar 2026 17:15:14 -0400 Subject: [PATCH 2/4] Update security section --- docs/enterprise/security.mdx | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-) diff --git a/docs/enterprise/security.mdx b/docs/enterprise/security.mdx index e5a8d8c..ff001fa 100644 --- a/docs/enterprise/security.mdx +++ b/docs/enterprise/security.mdx @@ -5,20 +5,22 @@ description: "Learn about LanceDB Enterprise security features and best practice icon: "shield-alt" --- -LanceDB Enterprise maintains the highest security standards with SOC 2 Type II and HIPAA compliance certifications. Our security framework ensures enterprise-grade protection for your data and workloads across all deployment models. +LanceDB Enterprise maintains high security standards with SOC 2 Type II, HIPAA, and GDPR compliance. Our security framework is designed to provide enterprise-grade protection for your data and workloads across deployment models. ## Security Certifications - **SOC 2 Type II**: Independent audit confirming our security controls and operational effectiveness - **HIPAA Compliance**: Certified to handle protected health information (PHI) in healthcare applications +- **GDPR Compliance**: Supports organizations with data privacy requirements under the General Data Protection Regulation - **Regular Audits**: Ongoing security assessments to maintain compliance standards -### Future Compliance +### Ongoing Compliance -Going forward, LanceDB will maintain SOC 2 Type II and HIPAA compliance by conducting continuous -audits to ensure our security practices remain aligned with industry standards and evolving -risks. Meanwhile, we are actively working on GDPR compliance. -[Contact us](mailto:contact@lancedb.com) to request a letter of engagement. +LanceDB maintains SOC 2 Type II, HIPAA, and GDPR compliance through ongoing audits and continuous improvement of our security practices as standards and risks evolve. + + +Visit the [LanceDB Trust Center](https://trust.lancedb.com/) to learn more about LanceDB's security posture, data privacy practices, and to request access to security documentation. + ## LanceDB Enterprise From a39561a41b032898c9d6cf96a4d9e9953e68658c Mon Sep 17 00:00:00 2001 From: prrao87 <35005448+prrao87@users.noreply.github.com> Date: Fri, 13 Mar 2026 17:17:35 -0400 Subject: [PATCH 3/4] Update title --- docs/enterprise/security.mdx | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/enterprise/security.mdx b/docs/enterprise/security.mdx index ff001fa..019219a 100644 --- a/docs/enterprise/security.mdx +++ b/docs/enterprise/security.mdx @@ -1,5 +1,5 @@ --- -title: "Security & compliance" +title: "Security & Compliance" sidebarTitle: "Security" description: "Learn about LanceDB Enterprise security features and best practices." icon: "shield-alt" From 275efbcc4820a72d5e5404016d5770a9bc6490c5 Mon Sep 17 00:00:00 2001 From: prrao87 <35005448+prrao87@users.noreply.github.com> Date: Fri, 13 Mar 2026 17:18:08 -0400 Subject: [PATCH 4/4] Fix broken links --- docs/enterprise/index.mdx | 2 +- docs/search/full-text-search.mdx | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/docs/enterprise/index.mdx b/docs/enterprise/index.mdx index b86689f..225cc04 100644 --- a/docs/enterprise/index.mdx +++ b/docs/enterprise/index.mdx @@ -80,7 +80,7 @@ With Lance OSS every query fetches data from S3, GCS, or Azure Blob. Each round LanceDB Enterprise uses NVMe SSDs as a hybrid cache, before the data store is even accessed. The first read fills the cache, and subsequent reads come from the local disk and return in tens of milliseconds. Parallel chunked reads further reduce tail latency. This gap matters when the application serves interactive dashboards or real-time recommendations. -Read More: [LanceDB Enterprise Performance](/enterprise/benchmarks/) +Read More: [LanceDB Enterprise Performance](/enterprise/performance/) ### Throughput of search queries diff --git a/docs/search/full-text-search.mdx b/docs/search/full-text-search.mdx index 2d0d28b..16064ed 100644 --- a/docs/search/full-text-search.mdx +++ b/docs/search/full-text-search.mdx @@ -697,7 +697,7 @@ const multiMatchBoostingResults = await table.query() - Create full-text search indices on text columns that will be frequently searched - For hybrid search combining text and vectors, see our [hybrid search guide](/search/hybrid-search/) -- For performance benchmarks, check our [benchmark results](/enterprise/benchmarks/) +- For performance benchmarks, check our [benchmark results](/enterprise/performance/) - For complex queries, use SQL to combine FTS with other filter conditions