{"id":7996,"date":"2024-05-21T08:30:51","date_gmt":"2024-05-21T15:30:51","guid":{"rendered":"https:\/\/devblogs.microsoft.com\/cosmosdb\/?p=7996"},"modified":"2024-09-10T14:39:31","modified_gmt":"2024-09-10T21:39:31","slug":"introducing-vector-database-capabilities-in-azure-cosmos-db-for-nosql","status":"publish","type":"post","link":"https:\/\/devblogs.microsoft.com\/cosmosdb\/introducing-vector-database-capabilities-in-azure-cosmos-db-for-nosql\/","title":{"rendered":"Introducing vector database capabilities in Azure Cosmos DB for NoSQL (Public Preview)"},"content":{"rendered":"<div class=\"content user-select-text\" tabindex=\"0\" aria-description=\"\" aria-label=\"Sent by Copilot: # Announcing Cosmos DB for NoSQL Vector Search with DiskANN We are thrilled to announce that Azure Cosmos DB for NoSQL has become the world\u2019s first full-featured serverless database with vector search capabilities. This groundbreaking development allows customers to store their data and vectors together, eliminating the need for a separate vector database. The result is improved consistency, ensured synchronization between vectors and data, and a reduction in the complexity and costs of AI application architectures. ## Unveiling the Benefits of Cosmos DB + DiskANN ### Scalability With the DiskANN Index stored on SSD and efficient graph construction, we can enable larger sizes compared to in-memory indexes like HNSW. By storing vectors on disk versus in-memory on a VM, you can leverage one of the most important features of Azure Cosmos DB - essentially infinite scale. You can grow to sizes that are drastically larger, with indexes in the 100\u2019s of billions of vectors and trillions of data points. ### Low Latency Long-range edges help search convergence faster and minimize the number of hops in the graph to reduce disk latency. DiskANN eliminates latency by combining an in-memory index of compressed vectors with a full-fidelity graph on fast SSDs. The small, in-memory index provides fast access for very low latency with an average query latency of less than 5ms. ### High Accuracy We link nodes to directionally diverse neighbors to improve recall. After the search operation, the results are re-ranked using the full-precision vectors themselves, providing very high accuracy. ### Low Cost Resource consumption is reduced by storing the full-fidelity graph on SSDs rather than in memory. And because this full-fidelity graph is on SSDs rather than entirely in-memory, it is much less expensive and can grow to sizes that are drastically larger, with indexes in the 100\u2019s of billions of vectors and trillions of data points. ### Robust to Data Changes Unlike HNSW, the DiskANN index itself is uniquely capable of supporting transactional workloads and does not degrade over time with high volumes of inserts, updates, or deletes. This is a differentiator among typical vector databases in the market today, which are built using HNSW and other less robust methods. ## Additional Benefits of Azure Cosmos DB - Serverless or provisioned throughput - Instant &amp; dynamic autoscale - Globally-replicated - Industry-leading 99.999% SLA Stay tuned for more updates as we continue to innovate and enhance our offerings to serve you better.\">\n<div class=\"ac-container ac-adaptiveCard\">\n<div class=\"ac-textBlock\">\n<p>\u00a0<span class=\"s1\">We are excited to announce that native vector indexing and search in Azure Cosmos DB for NoSQL is now available in preview! Azure Cosmos DB is the world\u2019s first full-featured serverless database with vector search and features multiple vector index options from flat (exact), quantized flat, and a new DiskANN-based index. DiskANN is a suite of highly scalable, accurate, and cost-effective approximate nearest neighbor (ANN) algorithms, developed at Microsoft Research, for low-latency and cost-effective vector search at any scale.<\/span><\/p>\n<\/div>\n<p class=\"p1\"><span class=\"s1\">You can take advantage of Azure Cosmos DB&#8217;s rich features such as a NoSQL query syntax to combine vector search with query filters that can increase the relevancy and accuracy of your vectors searches. You&#8217;ll also get all the benefits of Azure Cosmos DB&#8217;s flexibility, instant autoscale, 99.999% SLA, geo-replication, and more! Store your data and vectors together, eliminating the need to store vectors in a separate vector database and realize\u00a0improved consistency, synchronization between vectors and data, and a reduction in the complexity and costs of AI applications.<\/span><\/p>\n<p style=\"text-align: center;\"><iframe src=\"\/\/www.youtube.com\/embed\/u0ulNfZAxgc\" width=\"560\" height=\"314\" allowfullscreen=\"allowfullscreen\"><\/iframe><\/p>\n<div class=\"ac-textBlock\">\n<p><span style=\"color: inherit; font-family: arial, helvetica, sans-serif; font-size: 3rem; text-align: var(--bs-body-text-align);\">What is DiskANN?<\/span><\/p>\n<\/div>\n<\/div>\n<div class=\"content user-select-text\" tabindex=\"0\" aria-description=\"\" aria-label=\"Sent by Copilot: Sure, here's a brief description of DiskANN in four bullet points: - DiskANN is a **scalable** suite of approximate nearest neighbor search algorithms designed for **large-scale vector search**. - It supports **real-time changes** and simple filters, making it adaptable to dynamic data environments. - DiskANN can index and serve **5-10 times more points per node** compared to other methods like Hierarchical Navigable Small Worlds (HNSW). - It offers **high recall**, **low query latency**, and **high density**, even for billion-point datasets, making it a powerful tool for handling large volumes of data.\">\n<div class=\"ac-container ac-adaptiveCard\">\n<div class=\"ac-textBlock\">\n<p>DiskANN is a\u00a0suite of <strong>scalable <\/strong>approximate nearest neighbor search algorithms designed for <strong>efficient vector search at any scale<\/strong>. It offers <strong>high recall, high queries per second (QPS), and<\/strong>\u00a0<strong>low query latency <\/strong>even for billion-point datasets. This makes it it a powerful tool for handling large volumes of data. <a href=\"https:\/\/www.microsoft.com\/en-us\/research\/project\/project-akupara-approximate-nearest-neighbor-search-for-large-scale-semantic-search\/\" target=\"_blank\" rel=\"noopener\">Learn more about DiskANN from Microsoft.<\/a><\/p>\n<\/div>\n<p><a href=\"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/DiskANN.png\"><img decoding=\"async\" class=\"aligncenter wp-image-8168\" src=\"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/DiskANN.png\" alt=\"Image DiskANN\" width=\"1140\" height=\"641\" srcset=\"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/DiskANN.png 2500w, https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/DiskANN-300x169.png 300w, https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/DiskANN-1024x576.png 1024w, https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/DiskANN-768x432.png 768w, https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/DiskANN-1536x864.png 1536w, https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/DiskANN-2048x1152.png 2048w\" sizes=\"(max-width: 1140px) 100vw, 1140px\" \/><\/a><\/p>\n<ul>\n<li>DiskANN is a graph-based indexing and search system that performs fast and accurate approximate nearest neighbor (ANN) search at <strong>any-scale<\/strong>.<\/li>\n<li>It primarily uses an SSD-based index to scale to an order of magnitude more points compared to in-memory indices, while still retaining <strong>high QPS and low latency.<\/strong><\/li>\n<li>Quantized (compressed) vectors are kept in memory, and DiskANN balances interactions between the two to offer low latency and high accuracy.<\/li>\n<li>DiskANN is based on a novel graph index called <strong>Vamana<\/strong> that is more versatile than existing graph indices by maintaining accuracy despite many insertions, modifications, and deletions, without the need for expensive index rebuilds.<\/li>\n<\/ul>\n<\/div>\n<div class=\"ac-container ac-adaptiveCard\">\n<p><span style=\"font-family: arial, helvetica, sans-serif; color: inherit; font-size: 3rem; text-align: var(--bs-body-text-align);\">The DiskANN Advantage<\/span><\/p>\n<\/div>\n<\/div>\n<\/div>\n<div class=\"content user-select-text\" tabindex=\"0\" aria-description=\"\" aria-label=\"Sent by Copilot: # Announcing Cosmos DB for NoSQL Vector Search with DiskANN We are thrilled to announce that Azure Cosmos DB for NoSQL has become the world\u2019s first full-featured serverless database with vector search capabilities. This groundbreaking development allows customers to store their data and vectors together, eliminating the need for a separate vector database. The result is improved consistency, ensured synchronization between vectors and data, and a reduction in the complexity and costs of AI application architectures. ## Unveiling the Benefits of Cosmos DB + DiskANN ### Scalability With the DiskANN Index stored on SSD and efficient graph construction, we can enable larger sizes compared to in-memory indexes like HNSW. By storing vectors on disk versus in-memory on a VM, you can leverage one of the most important features of Azure Cosmos DB - essentially infinite scale. You can grow to sizes that are drastically larger, with indexes in the 100\u2019s of billions of vectors and trillions of data points. ### Low Latency Long-range edges help search convergence faster and minimize the number of hops in the graph to reduce disk latency. DiskANN eliminates latency by combining an in-memory index of compressed vectors with a full-fidelity graph on fast SSDs. The small, in-memory index provides fast access for very low latency with an average query latency of less than 5ms. ### High Accuracy We link nodes to directionally diverse neighbors to improve recall. After the search operation, the results are re-ranked using the full-precision vectors themselves, providing very high accuracy. ### Low Cost Resource consumption is reduced by storing the full-fidelity graph on SSDs rather than in memory. And because this full-fidelity graph is on SSDs rather than entirely in-memory, it is much less expensive and can grow to sizes that are drastically larger, with indexes in the 100\u2019s of billions of vectors and trillions of data points. ### Robust to Data Changes Unlike HNSW, the DiskANN index itself is uniquely capable of supporting transactional workloads and does not degrade over time with high volumes of inserts, updates, or deletes. This is a differentiator among typical vector databases in the market today, which are built using HNSW and other less robust methods. ## Additional Benefits of Azure Cosmos DB - Serverless or provisioned throughput - Instant &amp; dynamic autoscale - Globally-replicated - Industry-leading 99.999% SLA Stay tuned for more updates as we continue to innovate and enhance our offerings to serve you better.\">\n<div class=\"ac-container ac-adaptiveCard\">\n<div class=\"ac-textBlock\">\n<p><strong><span style=\"font-family: arial, helvetica, sans-serif;\">Scalability<\/span><\/strong><\/p>\n<ul>\n<li>DiskANN vector indexes are stored on high-speed SSDs, while compressed vectors are stored in memory.\u00a0 This reduces memory-footprint of the vector index, enabling planet-sized scalability for vector search scenarios.<\/li>\n<\/ul>\n<p><strong><span style=\"font-family: arial, helvetica, sans-serif;\">Low Latency<\/span><\/strong><\/p>\n<ul>\n<li>The DiskANN graph index construction makes it very efficient during search, minimizing the number of SSD reads to achieve high throughput and low latency.<\/li>\n<\/ul>\n<p><strong><span style=\"font-family: arial, helvetica, sans-serif;\">High Accuracy<\/span><\/strong><\/p>\n<ul>\n<li>During index construction, nodes in the graph are connected to diverse neighbors to improve recall. After the search operation, the results are re-ranked using the full-precision vectors providing high accuracy.<\/li>\n<\/ul>\n<p><strong><span style=\"font-family: arial, helvetica, sans-serif;\">Low Cost<\/span><\/strong><\/p>\n<ul>\n<li>Because the quantized vectors are stored in memory and the full-precision graph is stored on SSDs, it&#8217;s much less expensive to maintain and operate DiskANN-based indexes. This results in lower RU costs for your vector search queries.<\/li>\n<\/ul>\n<p><strong><span style=\"font-family: arial, helvetica, sans-serif;\">Robust to Insertions, Deletions, and Modifications<\/span><\/strong><\/p>\n<ul>\n<li>The DiskANN graph index is capable of supporting transactional workloads and does not degrade over time with high volumes of inserts, updates, or deletes. This is a differentiator among typical vector databases in the market today, which are built using HNSW and other less robust methods that require computationally expensive full index rebuilds to maintain accuracy.<\/li>\n<\/ul>\n<p>The benefits of DiskANN, combined with the instant &amp; dynamic autoscale, global replication, and industry leading 99.999% SLA of Azure Cosmos DB make for an unparalleled database for managing both your operational and vector data workloads.<\/p>\n<\/div>\n<h2><span style=\"font-family: arial, helvetica, sans-serif;\">What vector index options are available?<\/span><\/h2>\n<\/div>\n<p>There are multiple types of vector index policies that can be defined for a Cosmos DB collection. <a href=\"https:\/\/aka.ms\/CosmosDBVectorIndexing\" target=\"_blank\" rel=\"noopener\">Learn more about vector indexing in Azure Cosmos DB<\/a><\/p>\n<\/div>\n<ul>\n<li><strong>Flat index<\/strong> is an exact (sometimes called brute-force) approach to vector indexing. The vectors are placed on the Azure Cosmos DB index and referenced for efficient lookup. This may be a good option to use in scenarios where 100% accuracy of vector searches is required, and both the dimensionality of the vectors is small.<\/li>\n<li><strong>Quantized Flat index<\/strong> is also an exact approach, but the vectors are quantized (compressed) before being added to the Azure Cosmos DB index. This is very efficient and uses the same quantization method featured in DiskANN. <strong>Note:<\/strong>\u00a0Quantized Flat index needs least 1,000 vectors to ensure quantization accuracy.<\/li>\n<li><strong>DiskANN index <\/strong>enables approximate nearest neighbors (ANN) search at scale, with efficiencies that reduce RU cost and latency. This is extremely efficient and low-cost, especially when you expect to scale to larger scenarios.<\/li>\n<\/ul>\n<p>This table provides a good guide for the different index types and their strengths:<\/p>\n<table class=\" aligncenter\" style=\"height: 333px; width: 96.782%;\" width=\"82.3531%\">\n<tbody>\n<tr style=\"height: 47px;\">\n<td style=\"width: 8.85509%; height: 47px;\" width=\"108\"><strong>Type<\/strong><\/td>\n<td style=\"width: 16.2179%; height: 47px;\" width=\"150\"><strong>Description<\/strong><\/td>\n<td style=\"width: 39.5714%; height: 47px;\" width=\"144\"><strong>When to use it?<\/strong><\/td>\n<td style=\"width: 9.12722%; text-align: center; height: 47px;\" width=\"58\"><strong>Max # of dimensions<\/strong><\/td>\n<td style=\"width: 8.1126%; text-align: center; height: 47px;\" width=\"55\"><strong>RU Cost <\/strong><\/td>\n<td style=\"width: 7.36952%; text-align: center; height: 47px;\" width=\"55\"><strong>Speed<\/strong><\/td>\n<td style=\"width: 22.4901%; text-align: center; height: 47px;\" width=\"53\"><strong>Accuracy<\/strong><\/td>\n<\/tr>\n<tr style=\"height: 103px;\">\n<td style=\"width: 8.85509%; height: 103px;\" width=\"108\">Flat<\/td>\n<td style=\"width: 16.2179%; height: 103px;\" width=\"150\">Exact search on full vectors<\/td>\n<td style=\"width: 39.5714%; text-align: left; height: 103px;\" width=\"144\">\n<ul>\n<li>100% accuracy is required<\/li>\n<li><span style=\"font-family: inherit; font-size: inherit;\">The container has &lt;10k vectors<\/span><\/li>\n<li><span style=\"font-family: inherit; font-size: inherit;\">Vectors are small in size<\/span><\/li>\n<\/ul>\n<\/td>\n<td style=\"width: 9.12722%; text-align: center; height: 103px;\" width=\"58\">505<\/td>\n<td style=\"width: 8.1126%; text-align: center; height: 103px;\" width=\"55\">High<\/td>\n<td style=\"width: 7.36952%; text-align: center; height: 103px;\" width=\"55\">Slow<\/td>\n<td style=\"width: 22.4901%; text-align: center; height: 103px;\" width=\"53\">Highest<\/td>\n<\/tr>\n<tr style=\"height: 103px;\">\n<td style=\"width: 8.85509%; height: 103px;\" width=\"108\">Quantized Flat<\/td>\n<td style=\"width: 16.2179%; height: 103px;\" width=\"150\">Exact search on quantized vectors.<\/td>\n<td style=\"width: 39.5714%; text-align: left; height: 103px;\" width=\"144\">\n<ul>\n<li>The container has &gt;1,000 vectors\n<span style=\"font-family: inherit; font-size: inherit;\">The scenario is scoped to a small number of vectors, generally under 100k.<\/span><\/li>\n<\/ul>\n<\/td>\n<td style=\"width: 9.12722%; text-align: center; height: 103px;\" width=\"58\">4096<\/td>\n<td style=\"width: 8.1126%; text-align: center; height: 103px;\" width=\"55\">Medium<\/td>\n<td style=\"width: 7.36952%; text-align: center; height: 103px;\" width=\"55\">Medium<\/td>\n<td style=\"width: 22.4901%; text-align: center; height: 103px;\" width=\"53\">High<\/td>\n<\/tr>\n<tr style=\"height: 79px;\">\n<td style=\"width: 8.85509%; height: 79px;\" width=\"108\">DiskANN<\/td>\n<td style=\"width: 16.2179%; height: 79px;\" width=\"150\">Fast, approximate search at any scale.<\/td>\n<td style=\"width: 39.5714%; text-align: left; height: 79px;\" width=\"144\">\n<ul>\n<li>Any scenario where scale and efficiency are needed<\/li>\n<\/ul>\n<\/td>\n<td style=\"width: 9.12722%; text-align: center; height: 79px;\" width=\"58\">4096<\/td>\n<td style=\"width: 8.1126%; text-align: center; height: 79px;\" width=\"55\">Lowest<\/td>\n<td style=\"width: 7.36952%; text-align: center; height: 79px;\" width=\"55\">Fast<\/td>\n<td style=\"width: 22.4901%; text-align: center; height: 79px;\" width=\"53\">High<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<div class=\"content user-select-text\" tabindex=\"0\" aria-description=\"\" aria-label=\"Sent by Copilot: # Announcing Cosmos DB for NoSQL Vector Search with DiskANN We are thrilled to announce that Azure Cosmos DB for NoSQL has become the world\u2019s first full-featured serverless database with vector search capabilities. This groundbreaking development allows customers to store their data and vectors together, eliminating the need for a separate vector database. The result is improved consistency, ensured synchronization between vectors and data, and a reduction in the complexity and costs of AI application architectures. ## Unveiling the Benefits of Cosmos DB + DiskANN ### Scalability With the DiskANN Index stored on SSD and efficient graph construction, we can enable larger sizes compared to in-memory indexes like HNSW. By storing vectors on disk versus in-memory on a VM, you can leverage one of the most important features of Azure Cosmos DB - essentially infinite scale. You can grow to sizes that are drastically larger, with indexes in the 100\u2019s of billions of vectors and trillions of data points. ### Low Latency Long-range edges help search convergence faster and minimize the number of hops in the graph to reduce disk latency. DiskANN eliminates latency by combining an in-memory index of compressed vectors with a full-fidelity graph on fast SSDs. The small, in-memory index provides fast access for very low latency with an average query latency of less than 5ms. ### High Accuracy We link nodes to directionally diverse neighbors to improve recall. After the search operation, the results are re-ranked using the full-precision vectors themselves, providing very high accuracy. ### Low Cost Resource consumption is reduced by storing the full-fidelity graph on SSDs rather than in memory. And because this full-fidelity graph is on SSDs rather than entirely in-memory, it is much less expensive and can grow to sizes that are drastically larger, with indexes in the 100\u2019s of billions of vectors and trillions of data points. ### Robust to Data Changes Unlike HNSW, the DiskANN index itself is uniquely capable of supporting transactional workloads and does not degrade over time with high volumes of inserts, updates, or deletes. This is a differentiator among typical vector databases in the market today, which are built using HNSW and other less robust methods. ## Additional Benefits of Azure Cosmos DB - Serverless or provisioned throughput - Instant &amp; dynamic autoscale - Globally-replicated - Industry-leading 99.999% SLA Stay tuned for more updates as we continue to innovate and enhance our offerings to serve you better.\">\n<p>&nbsp;<\/p>\n<p>&nbsp;<\/p>\n<p><span style=\"color: inherit; font-family: arial, helvetica, sans-serif; font-size: 3rem; text-align: var(--bs-body-text-align);\">Enroll in the Vector Search Preview<\/span><\/p>\n<div class=\"ac-container ac-adaptiveCard\">\n<div class=\"ac-textBlock\">\n<p>Vector search in Azure Cosmos DB for NoSQL is a preview feature and requires enrollment via the <em>Features<\/em> page of your Azure Cosmos DB resource . Follow the below steps to register:<\/p>\n<p><a href=\"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/Screenshot-2024-05-20-at-4.34.06\u202fPM.png\"><img decoding=\"async\" class=\"aligncenter size-full wp-image-8167\" src=\"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/Screenshot-2024-05-20-at-4.34.06\u202fPM.png\" alt=\"Image Screenshot 2024 05 20 at 4 34 06 PM\" width=\"1914\" height=\"743\" srcset=\"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/Screenshot-2024-05-20-at-4.34.06\u202fPM.png 1914w, https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/Screenshot-2024-05-20-at-4.34.06\u202fPM-300x116.png 300w, https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/Screenshot-2024-05-20-at-4.34.06\u202fPM-1024x398.png 1024w, https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/Screenshot-2024-05-20-at-4.34.06\u202fPM-768x298.png 768w, https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-content\/uploads\/sites\/52\/2024\/05\/Screenshot-2024-05-20-at-4.34.06\u202fPM-1536x596.png 1536w\" sizes=\"(max-width: 1914px) 100vw, 1914px\" \/><\/a>\n1. Navigate to your Azure Cosmos DB for NoSQL resource page.<\/p>\n<p>2. Select the &#8220;Features&#8221; pane under the &#8220;Settings&#8221; menu.<\/p>\n<p>3. Select \u201cVector Search in Azure Cosmos DB for NoSQL\u201d.<\/p>\n<p>4. Read the description of the feature and confirm you want to enroll in the preview.<\/p>\n<p>5. Select &#8220;Enable&#8221; to enroll in the preview.<\/p>\n<p>&nbsp;<\/p>\n<\/div>\n<\/div>\n<h2><span style=\"font-family: arial, helvetica, sans-serif;\">Next Steps<\/span><\/h2>\n<p>The integration of vector search capabilities into Azure Cosmos DB for NoSQL marks a significant advancement in database technology, offering unparalleled scalability, efficiency, and accuracy. With the introduction of DiskANN and other vector indexing options, Azure Cosmos DB provides robust solutions for managing large-scale vector data alongside your operational data. Enroll in the Vector Search Preview today and explore the future of AI-driven applications with the powerful features of Azure Cosmos DB.<\/p>\n<ul>\n<li><a href=\"https:\/\/aka.ms\/CosmosVectorSearch\" target=\"_blank\" rel=\"noopener\">Learn more about Vector Search Preview in Azure Cosmos DB.\u00a0<\/a><\/li>\n<li><a href=\"https:\/\/www.youtube.com\/watch?v=MlMPIYONvfQ\">Learn about DiskANN &amp; Azure Cosmos DB NoSQL from Microsoft Mechanics<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/microsoft\/AzureDataRetrievalAugmentedGenerationSamples\/\" target=\"_blank\" rel=\"noopener\">Azure Cosmos DB in Semantic Kernel<\/a><\/li>\n<li><a href=\"https:\/\/python.langchain.com\/v0.1\/docs\/integrations\/vectorstores\/azure_cosmos_db\/\" target=\"_blank\" rel=\"noopener\">Azure Cosmos DB | \ud83e\udd9c\ufe0f\ud83d\udd17 LangChain<\/a><\/li>\n<li><a href=\"https:\/\/learn.microsoft.com\/en-us\/semantic-kernel\/memories\/vector-db\" target=\"_blank\" rel=\"noopener\">Vector Database | Microsoft Learn<\/a><\/li>\n<li><a href=\"https:\/\/www.microsoft.com\/research\/publication\/diskann-fast-accurate-billion-point-nearest-neighbor-search-on-a-single-node\/\" target=\"_blank\" rel=\"noopener\">DiskANN &#8211; Microsoft Research<\/a><\/li>\n<\/ul>\n<h2><span style=\"font-family: arial, helvetica, sans-serif;\">About Azure Cosmos DB<\/span><\/h2>\n<p>Azure Cosmos DB is a fully managed and serverless distributed database for modern app development, with SLA-backed speed and availability, automatic and instant scalability, and support for open-source PostgreSQL, MongoDB and Apache Cassandra.\u00a0<a href=\"https:\/\/cosmos.azure.com\/try\/\" target=\"_blank\" rel=\"noopener\">Try Azure Cosmos DB for free here<\/a>. To stay in the loop on Azure Cosmos DB updates, follow us on\u00a0<a href=\"https:\/\/twitter.com\/AzureCosmosDB\" target=\"_blank\" rel=\"noopener\">X<\/a>,\u00a0<a href=\"https:\/\/aka.ms\/AzureCosmosDBYouTube\" target=\"_blank\" rel=\"noopener\">YouTube<\/a>, and\u00a0<a href=\"https:\/\/www.linkedin.com\/company\/azure-cosmos-db\/\" target=\"_blank\" rel=\"noopener\">LinkedIn<\/a>.<\/p>\n<p>To quickly build your first database, watch our\u00a0<a href=\"https:\/\/youtube.com\/playlist?list=PLmamF3YkHLoLLGUtSoxmUkORcWaTyHlXp\" target=\"_blank\" rel=\"noopener\">Get Started videos<\/a>\u00a0on YouTube and explore ways to\u00a0<a href=\"https:\/\/docs.microsoft.com\/azure\/cosmos-db\/optimize-dev-test\" target=\"_blank\" rel=\"noopener\">dev\/test free.<\/a><\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":"<p>\u00a0We are excited to announce that native vector indexing and search in Azure Cosmos DB for NoSQL is now available in preview! Azure Cosmos DB is the world\u2019s first full-featured serverless database with vector search and features multiple vector index options from flat (exact), quantized flat, and a new DiskANN-based index. DiskANN is a suite [&hellip;]<\/p>\n","protected":false},"author":118435,"featured_media":8103,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[1610,14],"tags":[1236,1121,1866,1867,1868],"class_list":["post-7996","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai","category-core-sql-api","tag-ai","tag-msbuild","tag-vector-database","tag-vector-db","tag-vector-search"],"acf":[],"blog_post_summary":"<p>\u00a0We are excited to announce that native vector indexing and search in Azure Cosmos DB for NoSQL is now available in preview! Azure Cosmos DB is the world\u2019s first full-featured serverless database with vector search and features multiple vector index options from flat (exact), quantized flat, and a new DiskANN-based index. DiskANN is a suite [&hellip;]<\/p>\n","_links":{"self":[{"href":"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-json\/wp\/v2\/posts\/7996","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-json\/wp\/v2\/users\/118435"}],"replies":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-json\/wp\/v2\/comments?post=7996"}],"version-history":[{"count":0,"href":"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-json\/wp\/v2\/posts\/7996\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-json\/wp\/v2\/media\/8103"}],"wp:attachment":[{"href":"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-json\/wp\/v2\/media?parent=7996"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-json\/wp\/v2\/categories?post=7996"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/cosmosdb\/wp-json\/wp\/v2\/tags?post=7996"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}