Presenters
Source
MongoDB Unlocks Search and Vector Search Everywhere: Your Data, Your Way! 🚀
Hey tech enthusiasts! 👋 Ever felt the power of a truly seamless search experience in your applications? What if you could harness that same magic, not just for text, but for understanding the meaning behind your data, all within your familiar MongoDB environment? Well, buckle up, because MongoDB is making it happen, and it’s more accessible than ever!
Genevieve Broadhead, a seasoned Solutions Architect and leader at MongoDB, recently shed light on how MongoDB is revolutionizing search capabilities, especially for its Enterprise Advanced users. The core message? Run anywhere, search everywhere.
The Evolution of Search: From Text to Meaning 💡
Remember the early days of MongoDB? The vision was clear: empower developers and make data astonishingly easy to work with. Over the years, MongoDB has consistently evolved, bringing innovations like Atlas, vector search, and performance enhancements. But the driving questions have always been: how has our interaction with data changed, and where do our customers want to run their platforms?
The Ubiquitous Need for Search 🌐
It dawned on us that search isn’t just for e-commerce websites anymore. From the sleek customer-facing interfaces to the crucial back-office applications, every interaction with data benefits from powerful search. Think about it: manually typing order numbers at a customer service desk? That’s a missed opportunity for efficiency! MongoDB recognized this, bringing full-text search to Atlas, powered by Apache Lucene, to make data instantly searchable.
AI’s Influence: Embracing Semantic Understanding 🤖
Then came the AI revolution. The buzz around agentic systems, AI, and semantic understanding of data led to the next big leap: Atlas Vector Search. Launched about two years ago, this innovation allows you to perform not just traditional queries and full-text searches, but also vector index searches on the same data, using the same API. This means you can query data based on its meaning, not just keywords.
Bridging the Gap: Search for All Your MongoDB Deployments 🌉
For a while, the cutting edge of these search features was primarily available on MongoDB Atlas, the fully managed cloud database. But what about the vast number of customers running MongoDB Enterprise Advanced or Enterprise Server on-premises or in their own cloud environments?
MongoDB heard the call! In September and October, they announced MongoDB Search and Vector Search on Enterprise Server. This is a game-changer, bringing that same powerful search experience to wherever you run your MongoDB, whether it’s community, Enterprise Advanced, or Atlas.
Why is this a Big Deal? 🤔
Genevieve highlighted three key reasons for bringing these advanced search capabilities to self-managed clusters:
- Run Search Workloads Anywhere 🏠: No matter where your MongoDB application resides today – on-premises, in your private cloud, or even on your laptop – you can now equip it with search and vector search capabilities. This empowers your developers immediately.
- Eliminate Synchronization Tax 🔄: Before, if you needed search and vector search, you often had to maintain separate search engines and vector databases. This meant complex data synchronization, leading to delays and increased architectural complexity. MongoDB’s integrated approach eliminates this hassle. When you add data, your indexes (both traditional and vector) update automatically. The result? Your data is searchable the moment it’s added, across all desired index types.
- Superior Developer Experience ✨: In today’s fast-paced development world, reducing friction is key. MongoDB’s unified API means developers can perform normal queries, full-text searches, and vector searches using the same query language (MQL) and API. This significantly lowers the barrier to entry for building AI-powered applications, allowing developers to leverage their existing skills without learning entirely new languages or technologies.
Empowering Builders: From Startups to Enterprises 💼
This expansion of search capabilities is a boon for various user segments:
For the Builders of Tomorrow (Startups) 🚀
Many successful software companies, like those in retail with Adobe and Commerce Tools, started small, often on a developer’s laptop. MongoDB wants to ensure that the best tools and architectures are available from day one. By bringing search and vector search to MongoDB Community and Enterprise Server, they’re empowering new ventures to build cutting-edge AI applications from the ground up. This also ensures consistency across development, staging, and production environments, catching issues earlier.
For Enterprises and Regulated Industries 🏦
Enterprises, especially in sectors like banking and finance, often have strict data residency and governance requirements that necessitate on-premises or hybrid cloud deployments. Previously, these organizations might have managed separate databases for their core data, search, and vector embeddings, leading to complex synchronization and reduced feature adoption.
Now, with MongoDB Search and Vector Search on Enterprise Server, these businesses can:
- Maintain Data Sovereignty: Keep sensitive data within their own data centers or chosen cloud environments, complying with regulations and data residency laws.
- Reduce Architectural Complexity: Consolidate their data infrastructure, eliminating the need for multiple disparate databases and their associated synchronization challenges.
- Future-Proof Their Architecture: Safeguard against future cloud exit strategies or architectural shifts, knowing they have a consistent, portable solution.
The “Run Anywhere” Strategy in Action 🌐
Genevieve emphasized that this move aligns perfectly with MongoDB’s “run anywhere” strategy. Whether you’re running MongoDB on AWS, GCP, Azure, on-premises, or even locally, you get a consistent, feature-rich experience. This flexibility is crucial for independent software vendors (ISVs) and multi-tenant SaaS companies who need to deploy their applications wherever their customers are.
Getting Started: It’s Easier Than You Think! 🛠️
Ready to dive in? Setting up search and vector search on your self-managed MongoDB clusters is surprisingly straightforward:
- Kubernetes Environment: You’ll need a Kubernetes environment. Your MongoDB cluster can be inside or outside Kubernetes, but the search and vector search nodes will reside within it.
- MongoDB Controller for Kubernetes: Install the latest version of the MongoDB Controller for Kubernetes to easily deploy resources.
- Custom Resource: Create a custom search resource via the Kubernetes operator.
- Deploy and Connect: Deploy the search nodes and connect them to your existing MongoDB cluster.
The architecture is designed for seamless integration. Whether your MongoDB
server is inside or outside Kubernetes, the MongoDB Controller orchestrates the
deployment of the search (T process) and vector search capabilities, allowing
you to manage everything through a single interface.
Creating Indexes and Queries Made Simple ✍️
Creating indexes is intuitive, mirroring familiar MongoDB commands. For vector indexes, you specify dimensions and similarity algorithms like cosine. For full-text search, you can index all fields or target specific ones.
The real magic happens when you query. Using the aggregation framework, you’ll
use new functions like $search for full-text queries and vector search
queries. The beauty is that you don’t need to learn entirely new query
languages. The syntax is familiar and human-readable.
For example, a full-text search query for “cookie” on the product_name field
looks like this:
db.collection.aggregate([
{
$search: {
index: "your_full_text_index_name",
text: {
query: "cookie",
path: "product_name",
},
},
},
]);
And a vector search query:
db.collection.aggregate([
{
$vectorSearch: {
index: "your_vector_index_name",
path: "your_embedding_path",
query: [
/* your embedding vector */
],
numCandidates: 100, // Example
limit: 5, // Example
},
},
]);
This unified approach puts powerful AI skills directly into the hands of your developers without the steep learning curve often associated with new technologies.
A Seamless Demo: Atlas to Localhost 💻
Genevieve showcased a compelling demo illustrating this seamless transition. A retail product inventory search application, initially running on MongoDB Atlas, was reconfigured to run locally on a laptop, connecting to a self-managed MongoDB instance. The result? Identical functionality, from semantic search for “probiotics” to fuzzy matching for a misspelled “cookie,” all running locally. This highlights the power of a consistent architecture across environments, a critical factor for cloud exit strategies and architectural flexibility.
The code for this demo, along with other industry-specific examples (financial services, insurance, healthcare), is available on GitHub and the MongoDB Industry Solutions Hub.
The Future is Here: Integrated, Powerful, and Accessible 🌐
MongoDB’s commitment to making data easy to work with has never been stronger. By bringing robust search and vector search capabilities to its Enterprise Advanced platform, MongoDB is empowering developers and enterprises alike to build smarter, more responsive applications. Whether you’re a startup founder with a groundbreaking idea or an enterprise navigating complex regulatory landscapes, MongoDB offers a unified, powerful, and flexible data platform to fuel your innovation.
So, go ahead, unlock the power of your data, and start building the future today! ✨