MongoDB deprecated the Atlas Data API as part of the broader shutdown of Atlas App Services. The end-of-life date of September 30, 2025 has passed, and if you depended on HTTP-based access to your MongoDB data - whether directly or through a low-code integration - you need an alternative.
The good news is there are several solid options. The challenge is that each one makes different trade-offs. Some are drop-in replacements that mimic the Atlas Data API request format. Others are full API platforms that go beyond what MongoDB ever offered. Some are MongoDB-only. Others support your entire database stack.
This post compares the leading alternatives. We'll be straightforward: DreamFactory is one of these options - a secure, self-hosted platform providing governed API access to any data source for enterprise apps and on-prem LLMs. We think it has a unique advantage for teams with multi-database environments. But we'll present each option honestly and let the comparison speak for itself.
Before comparing alternatives, it's worth understanding the full scope of what was removed. Most coverage focuses on the Data API, but MongoDB deprecated the entire Atlas App Services platform. That includes the Data API, Custom HTTPS Endpoints, Device Sync, Device SDKs (Realm), GraphQL API, Functions, Static Hosting, and Edge Server.
The Data API specifically was the HTTP-based CRUD interface that let you interact with MongoDB collections over HTTPS without installing database drivers. You sent a POST request with a JSON payload describing your query, and the API returned results. The key use cases it served were serverless and edge environments (Cloudflare Workers, AWS Lambda), low-code and no-code integrations (Azure Logic Apps, Zapier, Retool), mobile apps, and IoT devices.
MongoDB's deprecation page lists several recommended alternatives. But no single alternative covers every use case, and the right choice depends heavily on your stack.
Here's how the major options compare across the dimensions that matter most:
|
Solution |
Type |
MongoDB Support |
SQL Database Support |
Self-Hosted |
License |
Listed on MongoDB's Deprecation Page |
|---|---|---|---|---|---|---|
|
RESTHeart |
Open source server |
Native |
No |
Yes |
AGPL-3.0 |
Yes |
|
Delbridge Data API |
Drop-in replacement |
Native |
No |
Yes |
Open Source |
Yes |
|
Neurelo |
Cloud platform |
Yes |
PostgreSQL, MySQL |
Limited |
Proprietary |
Yes |
|
Hasura |
GraphQL engine |
Via connector |
Yes (many) |
Yes |
Apache 2.0 |
Yes |
|
FerretDB |
MongoDB-compatible DB |
Via PostgreSQL backend |
No |
Yes |
Apache 2.0 |
No |
|
FetchDB (Truto) |
Cloud API proxy |
Yes |
No |
No |
Proprietary |
No |
|
DIY (Express/FastAPI) |
Custom code |
Yes |
Whatever you build |
Yes |
N/A |
Yes (tutorials) |
|
DreamFactory |
Governed API |
Yes |
20+ databases |
Yes |
Apache 2.0 |
No |
Now let's look at each one in detail.
RESTHeart is a mature, MongoDB-native REST and GraphQL API server built by SoftInstigate. It has over 10 years of development history - this isn't a startup scrambling to capture the deprecation market. They've been building MongoDB API tooling since before Atlas Data API existed.
RESTHeart is listed on MongoDB's official deprecation page as a recommended alternative, and they've built a dedicated migration landing page for teams moving off the Data API.
The platform provides REST and GraphQL APIs for MongoDB out of the box, with support for MongoDB Atlas, Percona Server for MongoDB, AWS DocumentDB, and Azure Cosmos DB. It includes an access manager with role-based security, JSON Schema validation, change streams, and aggregation pipeline support.
Where RESTHeart fits well: MongoDB-only environments where you want a proven, self-hosted solution with deep MongoDB expertise behind it.
Where it falls short: The AGPL-3.0 license is the biggest consideration for enterprise buyers. AGPL requires that if you distribute the software as a network service, you must make your source code available under the same license. Many enterprise legal teams flag AGPL as a risk, and some organizations prohibit AGPL-licensed software entirely. SoftInstigate offers commercial licensing, but this adds cost and vendor dependency.
The other limitation is scope. RESTHeart is MongoDB-only. If your stack includes PostgreSQL, MySQL, SQL Server, or any other SQL database alongside MongoDB, you'll need a separate API solution for those databases. RESTHeart can technically work with PostgreSQL through a FerretDB translation layer, but that's a workaround - not native SQL support.
Delbridge has positioned themselves as the most direct drop-in replacement for Atlas Data API. They're listed on MongoDB's official deprecation page, and MongoDB published a blog post about them in July 2025. If there's a "MongoDB-approved" alternative, Delbridge is it.
The product is designed so migration is as simple as updating your endpoint URL and authentication logic. It supports full CRUD operations, native MQL (MongoDB Query Language), EJSON compatibility, and bulk operations - matching the Atlas Data API feature set closely enough that existing client code needs minimal changes.
Delbridge is primarily a consulting company with over 500 MongoDB migrations completed. The Data API product is an extension of that consulting business, which has pros and cons. On the plus side, they deeply understand MongoDB migration challenges. On the minus side, the product is secondary to the consulting revenue.
Where Delbridge fits well: Teams that need their existing Atlas Data API calls to keep working with the absolute minimum code changes. If your priority is "fix this fast," Delbridge is the fastest path.
Where it falls short: Like RESTHeart, Delbridge is MongoDB-only. If you need REST APIs for PostgreSQL, MySQL, SQL Server, or any other database, you need a second tool. The drop-in replacement approach also means you're adopting the same API design patterns that MongoDB deprecated - POST-for-everything, non-standard REST conventions. If you're going to invest in migration effort, there's an argument for migrating to a more standard REST API design rather than replicating the old one.
Neurelo is a cloud-based data API platform with AI-powered features. It supports MongoDB, PostgreSQL, and MySQL, and has published specific Atlas Data API migration guides for both REST and GraphQL endpoints. Neurelo is listed on MongoDB's official deprecation page.
The AI features are genuinely interesting: text-to-schema generation, AI-assisted query building, and natural language interfaces to your data. Their messaging has recently pivoted toward positioning themselves as an "Enterprise-Grade AI Database Platform for AI Agents" - suggesting they see the future of their business in AI/LLM data access rather than traditional API generation.
Neurelo also recently introduced "self-hosted gateways," though this capability is new and more limited than their cloud offering.
Where Neurelo fits well: Cloud-native teams who want AI-assisted development features and multi-database support (MongoDB + PostgreSQL + MySQL) without managing infrastructure.
Where it falls short: Neurelo has historically been cloud-only, meaning your API requests route through their infrastructure. For teams with compliance requirements - HIPAA, PCI-DSS, FedRAMP - this creates questions about data processing and residency. The new self-hosted gateway option exists but is still maturing.
There's also a practical concern about longevity. Neurelo is a VC-funded startup. The Fauna precedent is worth remembering here: Fauna positioned itself as a serverless database alternative, attracted developers looking for Atlas Data API replacements, and then shut down in 2025. This isn't to say Neurelo will follow the same path, but enterprise buyers making long-term platform bets should weigh this risk.
Hasura is a powerful GraphQL engine with broad database support. It's listed on MongoDB's deprecation page and supports PostgreSQL, MySQL, SQL Server, MongoDB (via a connector), BigQuery, and several other data sources.
Hasura is open source under the Apache 2.0 license with self-hosted deployment available. Enterprise features include role-based access control, SSO, observability, and performance monitoring.
The platform also provides REST endpoints through what Hasura calls "RESTified" endpoints - you define a GraphQL query and Hasura generates a REST endpoint that executes it. This is useful but functionally different from auto-generated REST APIs.
Where Hasura fits well: Teams that are already using GraphQL or are willing to adopt it. Hasura's GraphQL-first approach is excellent for frontend-heavy applications that benefit from flexible querying and real-time subscriptions. The multi-database support is genuine and production-tested.
Where it falls short: If your existing applications consume REST APIs - which is the case for most teams migrating from Atlas Data API - Hasura's REST support is a wrapper around GraphQL, not a native REST implementation. Your team needs to learn GraphQL to configure the system, even if the consuming applications use the RESTified endpoints. For teams specifically looking for auto-generated REST APIs from their database schema, Hasura adds a conceptual layer that may not be necessary.
FerretDB takes a fundamentally different approach. Rather than providing an API layer on top of MongoDB, FerretDB is a MongoDB-compatible database that uses PostgreSQL as the storage backend. They released a Data API in FerretDB v2 in February 2025 that directly replaces the Atlas Data API HTTP interface.
FerretDB is open source under the Apache 2.0 license and is genuinely interesting technology - it translates MongoDB wire protocol commands into PostgreSQL queries, giving you MongoDB compatibility on top of a mature relational database.
Where FerretDB fits well: Teams who want to migrate off MongoDB entirely and onto PostgreSQL while maintaining MongoDB-compatible application code. If your strategy is "exit MongoDB," FerretDB provides a bridge.
Where it falls short: This is fundamentally a migration off of MongoDB, not a replacement for the Atlas Data API on top of your existing MongoDB cluster. Your data moves from MongoDB to PostgreSQL. That's a significant architectural change, not a Data API replacement. If you want to keep using MongoDB Atlas and just need HTTP access to it, FerretDB isn't solving your actual problem.
FetchDB, built by Truto, is a cloud-based service specifically designed as a drop-in replacement for Atlas Data API. It was built because Truto's own engineering team was running Cloudflare Workers with MongoDB and got hit by the deprecation.
The product targets a specific niche: developers in serverless and edge environments (especially Cloudflare Workers) who need HTTP-based MongoDB access and want a quick fix.
Where FetchDB fits well: Cloudflare Workers developers who need a fast, cloud-hosted replacement and are comfortable with a newer service.
Where it falls short: It's MongoDB-only, cloud-only, and proprietary. There's no self-hosted option. As a newer, smaller service, it carries the same longevity questions as any startup-built alternative - though Truto is a broader company (unified API platform) and FetchDB is one product within that portfolio.
MongoDB's deprecation page includes tutorials for building your own API layer using Express.js, FastAPI, Spring Boot, and serverless functions (AWS Lambda, Azure Functions).
This approach gives you maximum flexibility and zero vendor dependency.
Where DIY fits well: Teams with strong backend engineering resources who want full control over their API design, security model, and deployment.
Where it falls short: You now own the security layer (authentication, authorization, rate limiting), the API documentation, monitoring, versioning, and long-term maintenance. For a single database with a small number of endpoints, this is manageable. For multiple databases or complex access control requirements, it becomes a full-time job.
There's also a common organizational pattern worth noting: a developer builds a custom API layer, documents it minimally, and then changes roles or leaves the company. The team that inherits the code doesn't fully understand it, and the API layer slowly becomes technical debt. We see this frequently in enterprises evaluating DreamFactory - they're replacing a custom API layer that someone built years ago and nobody wants to maintain.
DreamFactory is a secure, self-hosted platform providing governed API access to any data source for enterprise apps and on-prem LLMs. It auto-generates REST APIs for over 20 database types, including MongoDB, PostgreSQL, MySQL, SQL Server, Oracle, Snowflake, IBM DB2, and more.
When you connect a database to DreamFactory, it introspects your schema (or collections, in MongoDB's case) and generates a full set of REST API endpoints with OpenAPI/Swagger documentation. Each endpoint is secured by default with API key authentication and role-based access control.
DreamFactory is self-hosted - deployable via Docker, Kubernetes, or native Linux installers on your infrastructure, including VPC, on-premises, and air-gapped environments. The open source core is Apache 2.0 licensed, with a commercial enterprise edition that adds advanced security features (SAML SSO, LDAP/AD integration, advanced RBAC, audit logging, and identity passthrough).
Where DreamFactory fits well: Enterprise and mid-market teams that use MongoDB alongside SQL databases. This is DreamFactory's unique differentiator in the Atlas Data API replacement market: every other alternative on this list is either MongoDB-only or GraphQL-first. DreamFactory is the only platform that provides governed API access to MongoDB and your SQL databases in a single deployment with unified security.
Most enterprises don't run MongoDB in isolation. They have PostgreSQL for some applications, MySQL for others, SQL Server for legacy systems, maybe Snowflake for analytics. With DreamFactory, you get one API layer, one security model, one set of API keys, and one audit trail for all of them.
Additional capabilities include server-side scripting for custom business logic (V8 JavaScript, PHP, Python), stored procedure exposure via REST, file storage API support (S3, Azure Blob, local), and SOAP-to-REST conversion.
Where it falls short: DreamFactory is not a drop-in replacement for the Atlas Data API request format. If your existing code sends POST requests in the Atlas Data API format (/action/find, /action/insertOne, etc.), you'll need to update your client code to use DreamFactory's standard REST conventions (GET /api/v2/mongodb/_table/collection). This is a migration step, though the result is a more RESTful API design.
DreamFactory also uses a different approach to MongoDB than some competitors. It treats MongoDB collections as "tables" in its REST API, which provides consistency across database types but means some MongoDB-specific operations (complex aggregation pipelines, for example) may require server-side scripting rather than direct API exposure.
Here's a decision framework based on common scenarios:
You need the fastest possible fix with minimal code changes. Delbridge Data API mimics the Atlas Data API format, so your existing client code needs the fewest changes. Update the endpoint URL and auth, and you're back online.
You're a MongoDB-only shop and want a proven, self-hosted solution. RESTHeart has been doing this for over a decade and is listed on MongoDB's deprecation page. Just make sure your legal team is comfortable with AGPL, or budget for a commercial license.
You want GraphQL and support multiple databases. Hasura is the strongest GraphQL option. The learning curve is real, but the platform is mature and well-supported.
You want to leave MongoDB for PostgreSQL. FerretDB gives you a migration path that preserves MongoDB-compatible application code while moving your data to PostgreSQL.
You're cloud-native and want AI-powered features. Neurelo supports MongoDB, PostgreSQL, and MySQL with interesting AI capabilities. Evaluate the cloud-only trade-offs for your compliance requirements.
You run Cloudflare Workers and need this solved today. FetchDB was built for exactly your problem.
You need REST APIs for MongoDB AND SQL databases, with enterprise security and self-hosted deployment. This is DreamFactory's sweet spot. No other alternative on this list covers this combination.
What is the best drop-in replacement for the Atlas Data API?
If your priority is minimal code changes, Delbridge Data API mimics the Atlas Data API request format most closely. However, if you want a more comprehensive solution, DreamFactory provides governed API access to MongoDB and 20+ other databases - going well beyond what the Atlas Data API ever offered.
Which Atlas Data API alternative supports the most databases?
DreamFactory supports over 20 databases - MongoDB, PostgreSQL, MySQL, SQL Server, Oracle, Snowflake, IBM DB2, and more - in a single self-hosted deployment. No other alternative on this list covers this range. RESTHeart, Delbridge, and FetchDB are MongoDB-only.
Is DreamFactory open source?
The DreamFactory core is Apache 2.0 licensed - no AGPL restrictions. The enterprise edition adds advanced features like SAML SSO, LDAP/Active Directory integration, advanced RBAC, and audit logging. Both are self-hosted on your infrastructure.
Can I self-host my Atlas Data API replacement?
DreamFactory, RESTHeart, and Delbridge all support self-hosted deployment. Cloud-only options like Neurelo and FetchDB cannot be self-hosted, which may disqualify them for compliance-sensitive environments. DreamFactory deploys via Docker, Kubernetes, or native Linux installers.
How does DreamFactory compare to building a custom API with Express or FastAPI?
A DIY approach gives you full control but requires building authentication, authorization, rate limiting, documentation, and monitoring from scratch. DreamFactory auto-generates all of this - providing governed API access with role-based security, identity passthrough, and OpenAPI docs out of the box.
Whichever alternative you choose, the lesson from this deprecation is worth internalizing: own your API layer. When your API platform is controlled by your database vendor, you're one deprecation notice away from a scramble. This happened with Parse. It happened with Fauna. It happened with MongoDB.
An independent API layer - whether it's DreamFactory, Hasura, RESTHeart, or even a well-maintained custom solution - puts you in control. Your database is your database. Your API layer is yours. They can evolve independently, and no single vendor decision can break both at once.
If you'd like to see how DreamFactory handles your MongoDB and SQL database environment, we offer a free 30-minute demo where we can walk through your specific setup.