MongoDB Atlas Evolves to Support Generative AI Application Development

MongoDB today announced it is enhancing its cloud database, MongoDB Atlas, with a series of new products and features to make it easier for developers to build modern applications.

These include new generative artificial intelligence capabilities for MongoDB’s Vector Search capabilities that enable retrieval and personalization of highly relevant information, dedicated resource search nodes with large-scale search workloads, and search nodes for high-speed data streams. Includes stream processing.

The new MongoDB Atlas features were announced, along with many other updates, at the company’s annual developer conference, MongoDB.local NYC. Also new today are the MongoDB Relational Migrator, a tool that simplifies application migration, and Google Cloud and AI, aimed at accelerating the use of generative AI to build new types of applications. is a new partnership.

MongoDB is the creator of the document-oriented MongoDB database. This database is used for a wide range of data-intensive applications and is valued for its ability to store information in multiple different formats. MongoDB Atlas is a cloud-hosted version of that database.

The company explained that database choice is important for companies that want to leverage generative AI, the technology at the heart of next-generation chatbots such as OpenAI LP’s ChatGPT. The integrated, fully managed, flexible and scalable database makes life much easier for development teams, the company says.

Today’s update means MongoDB Atlas is poised to be its database, the company said. The updated Vector Search feature is key to building generative AI apps, as data must be stored in “vectors”, or geometric representations.

Generative AI models work by measuring similarities between vectors, probabilistically constructing sentences and images from prompts, and returning accurate search results with more context than traditional search engines. increase. MongoDB Atlas Vector Search enables enterprises to support new workloads such as semantic search, text-to-image search, and personalized recommendation systems.

MongoDB Atlas Vector Search will also enable existing generative AI models to be augmented with additional data to produce more accurate results for specific use cases and domains, the company said.

Constellation Research Inc. analyst Doug Heschen told SiliconANGLE that vector search is a key enabler for generative AI workloads. “Since vectors are geometric/numerical representations of semantically similar text, images, audio, and other content, they may be useful for training custom large-scale language models using your own data. is infinite,” he pointed out. “Vectors can help improve natural language queries, NL text/image/code generation, etc.”

Meanwhile, MongoDB Atlas Stream Processing is a key update that makes it easier to process streaming data and enable support for real-time applications. In Hengsheng’s view, this is the most important update for customers as low-latency workloads are becoming more prevalent in enterprises.

“MongoDB really needed to strengthen in this area to live up to its mandate as a developer data platform,” said the analyst. “Rival data platforms for analytics, such as Snowflake and Databricks, already support real-time needs, so MongoDB fills a competitive gap.

Time series collections also improve scalability, and the ability to modify such data after it is ingested makes it easier to handle time series workloads. This is important because time series databases typically cannot change data after it is created, even if an error occurs.

Our final update concerns the ability to tier and query data on Microsoft Azure using MongoDB Atlas Online Archive and Atlas Data Federation. This allows customers to tie their MongoDB databases to the most cost-effective cloud storage tier while maintaining the ability to query with higher performance.

MongoDB Chief Development Officer Ittycheria (pictured) said the new features were prioritized based on customer feedback. “We are further supporting customers running the largest and most demanding mission-critical workloads that require continued improvements in scalability and flexibility,” he said. .

Migration and consolidation

The new MongoDB Relational Migrator tool is designed to help developers who want to migrate their existing applications from legacy databases to MongoDB, promising to do so in a risk-free and cost-effective manner.

Meanwhile, a new partnership with Google Cloud involves the integration of MongoDB and Vertex AI. Vertex AI is a suite of tools for data scientists to build, automate, standardize, and manage machine learning projects, including generative AI models.

Ittycheria said the shift to generative AI begins first and foremost with developers, who are tasked with integrating the technology into a new class of business applications. “We want to democratize access to innovative technology so that every developer can build the next big thing,” said the CEO. “Our strategic partnership with Google Cloud makes it easier for organizations of all shapes and sizes to incorporate AI into their applications.”

Heschen said the integration with Google Cloud impacts MongoDB’s new vector search capabilities. “MongoDB can now provide data represented as vectors to drive training of language models at scale. VertexAI is one of many tools available for developing custom LLMs.” ‘, he explained.

new field

The series of updates continued with the announcement of MongoDB Atlas for Industries, touted as a new program to help organizations accelerate cloud adoption and modernization through industry-specific expertise and integrated solutions. According to MongoDB, the aim is to provide access to expert-led architecture design reviews, knowledge accelerators that provide more relevant training for developer teams, and partnerships to build solutions to industry-specific challenges. It is to offer to customers.

MongoDB Atlas for Industries will see the company launch its first-ever vertical offering for financial services customers, with more coming later this year in manufacturing, automotive, insurance, healthcare, and retail. I plan to

Similar to stream processing updates, MongoDB’s vertical support follows in the footsteps of its competitors: Amazon Web Services Inc., Google Cloud, Snowflake, and Databricks, Heschen noted. “MongoDB starts in one of the industries on everyone’s shortlist: financial services, because that’s where the most funding is expected,” he added. “Typical outcomes of vertical industry clouds are pre-built solutions and shortcuts to common use cases. It remains to be seen what effect it will have in terms of reducing the time to completion.”

Finally, the company announced a number of more general database updates, including expanded programming language support for MongoDB Atlas. This simplifies the deployment of resources to Amazon Web Services using infrastructure as code. There is also a new Kotlin Driver for MongoDB that enables building Kotlin-based applications, and a MongoDB Atlas Kubernetes Operator that simplifies the task of working with containerized applications.

“Everything MongoDB announces is designed to make MongoDB a more comprehensive and complete ‘developer data platform,’” said Hengsheng. “The more MongoDB can provide developers with all the tools they need, the more reliable the platform will be for developers and the organizations they work for.”

Photo: Silicon ANGLE