Why MongoDB Is Adding Native Analytics and What It Means for Developers

Six years ago, while writing for ZDNet, we wondered where MongoDB was headed. Over time the answer emerged: to become more scalable and support a wider range of applications. MongoDB added native search for content management, time‑series support for IoT, and change streams to help e‑commerce predict next actions.

MongoDB customers also need a cloud solution that matches development tools and is easy to adopt. The result is Atlas, a managed cloud service now accounting for 60% of MongoDB’s business.

Beyond that, analysis is a crucial piece.

1. Why Introduce Analytics?

Adding analytics instantly raises the usefulness of operational apps: manufacturers improve preventive maintenance, healthcare providers choose optimal care plans, and e‑commerce or gaming firms enhance user interaction and reduce churn. These decision‑optimizing analytics complement operational databases.

Combining analytics with transactional databases is not new—HTAP, translytical, and enhanced transactional databases are examples.

Cloud‑native’s separation of compute and storage offers a chance to blend operational processing with analytics without hurting performance. Recent efforts like Oracle MySQL HeatWave and Google AlloyDB illustrate this direction.

Hybrid databases typically add columnar tables for analytics while retaining relational structures for easy conversion. Introducing nested document models makes translation harder.

MongoDB should therefore have its own analytics, focusing on fast decision‑making rather than complex modeling.

2. A Gradual Journey

MongoDB has begun supporting analytics with visual charts and BI connectors, making it comparable to MySQL in Tableau and Qlik. Visualization is just the first step; deeper data relationships and “why” questions remain challenging.

MongoDB aims to boost competitiveness with analytics but won’t replace specialized solutions like Snowflake, Redshift, or Databricks. Its analytics target is application developers, not data analysts.

Atlas can reserve dedicated analytics nodes, and soon customers will choose compute instances optimized for analytics, with near‑real‑time replication.

Future plans include multi‑cloud instance choices, normative guides, and machine‑learning‑driven instance selection.

Atlas Serverless, previewed last year, will soon launch fully, benefiting analytics workloads that differ from transactional spikes.

3. Can MongoDB Connect to SQL?

Early resistance to SQL in MongoDB gave way to reasoned development.

This week MongoDB introduced a new interface for reading Atlas data: Atlas SQL, the first true SQL access for MongoDB, offering fine‑grained views that reflect JSON document richness rather than flattening data for Tableau.

SQL integration will evolve over years, requiring richer data‑warehouse options and support for upserts and other core operations.

Alongside Atlas SQL, a preview columnar index improves analytical query performance, with future automation based on access patterns, metadata, Bloom filters, and query planner enhancements.

Atlas Data Lake will provide unified views of JSON documents across clusters and cloud object storage, enabling broader federated queries.

4. Human‑Centric Focus

MongoDB has long been a developer favorite. Its JavaScript/JSON roots align with popular languages, but avoiding SQL limited access to a large talent pool. Embracing SQL can broaden its appeal.

While MongoDB still favors the document model, expanding its audience requires uniting developers and SQL users, simplifying workflows, and avoiding the need to move data to separate warehouses.

5. Not Aiming to Replace Data Warehouses

Data lakes or intelligent lakehouses remain distinct. Complex modeling stays in dedicated analytics systems, while supporting analytics inside operational databases enables inline, near‑real‑time processes.

MongoDB will cooperate with Snowflake, Databricks, etc., allowing models built in warehouses to feed back into transactional flows for automated anomaly detection.

Implementing such closed‑loop analytics involves change streams, triggers, and functions, but future releases aim to hide this complexity and provide simple, real‑time analytics options.