WEBINARHow to build smarter AI apps with Python and MongoDB. Register now >
NEWLearn MongoDB with expert tutorials and tips on our new Developer YouTube channel. Subscribe >
Blog home
arrow-left

Scaling with Confidence: Introducing Standardized Atlas Admin API Rate Limiting

March 3, 2026 ・ 3 min read

Managing a fleet of database clusters at scale requires precision, automation, and—most importantly—predictability. Today, MongoDB is excited to announce that standardized rate limiting for the Atlas Admin API v2 is now generally available (GA).

Whether you are a DevOps engineer orchestrating large-scale configuration updates or a developer building high-volume automation workflows, this update gives you the visibility and resilience needed to keep your Atlas environments running smoothly.

What’s new in the v2 update?

Previously, rate limits across different API endpoints were enforced, but not explicitly surfaced to clients. With this update, we’ve implemented a standardized system based on the token bucket algorithm. This approach provides a balance of strict operational boundaries and the flexibility to handle "bursty" traffic when you need it most.

How the Token Bucket works:

  • Capacity: Each "bucket" (endpoint set) has a maximum number of tokens representing the total requests you can send at once.

  • Consumption: Every API request consumes one token.

  • Refill: Tokens are added back at a fixed rate (refillRate) over a specific interval (refillDurationSeconds).

  • Bursting: You can exceed the steady refill rate as long as there are tokens remaining in the bucket.

Why this matters for your team

Reliability is a two-way street. Standardized rate limiting protects the Atlas ecosystem from surge loads while empowering you with the data to build smarter clients. This means you’ll get:

  • Predictable performance: No more guessing why a request failed. Clear limits ensure high availability across all your Atlas projects.

  • Real-time visibility: New API response headers tell you exactly where you stand in real time.

  • Resilient automation: With standardized error codes, your scripts can implement intelligent retry logic and exponential backoff, reducing the need for manual intervention.

For developers: navigation and headers

The Admin API now returns standardized headers to help you manage your request flow programmatically:

HeaderDescription
RateLimit-Limit The maximum capacity of the token bucket for that endpoint.
RateLimit-RemainingHow many requests you have left before the bucket is empty.
Retry-After(Returned on 429 errors) The number of seconds to wait before retrying.

Handling a "429 Too Many Requests"

If you exhaust your token bucket, Atlas will return a 429 status code with a detailed JSON body:

 

JSON

Pro tip: We recommend a retry strategy using exponential backoff with jitter. If the Retry-After header says 10 seconds, wait at least that long.

How to get started

Ready to audit your current usage? You can view the limits for your specific organization or project by hitting the new rateLimits endpoint (see documentation for the rateLimits API):

 

Bash

 

For most users, the default limits provide ample room for even complex automation. However, if your specific use case requires a higher ceiling, please contact support via the MongoDB Support Hub to adjust your specific rate limits

megaphone
Next Steps

Check out the updated Atlas Administration API documentation to dive deeper into the endpoint sets and specific limits.

MongoDB Resources
Atlas Learning Hub|Customer Case Studies|AI Learning Hub|Documentation|MongoDB University