Background:
For each MongoDB Altas project, the clusters built in that project are built using a "network container", which tells MongoDB Atlas which pool of IPs it can select from given a CIDR range for its underlying infrastructure.

AWS minimum imposed CIDR value/subnet size: /28
MongoDB minimum imposed CIDR value for clusters deployed to AWS: /24

Azure minimum imposed CIDR value/subnet size: /29
MongoDB minimum imposed CIDR value for clusters deployed to Azure: /21

For customers who elect connecting to MongoDB Atlas via VPC peering, they know that IP space is a scarce resource. Reserving /24 or /21 IP space from IPAM does not scale when your project footprint in MongoDB Atlas grows sufficiently large.

Attempting to combine projects into a single cohesive project for utilizing the larger IP space (as MongoDB likely intended) does not scale either. There is a soft cap on the amount of clusters which can be deployed to a project (25). The cap can be raised but concerns with monitoring, backups, and visibility into all clusters in the project becomes a concern, as well as further complicating billing and RBAC management in larger orgs.

For customers with smaller deployments, like simple 3 node replica sets in AWS, MongoDB asking us to allocate a bare minimum of /24 (255 IPs) when only 3 IPs are needed, thats a ton of overhead on IPAM that goes to waste unnecessarily. The problem is even more exaggerated for 3 node replica sets in Azure, where the bare minimum network container size Atlas is /21 resulting in over 2000+ IPs going to waste.

Proposed solution:
MongoDB raise the minimum required network container size to /28 for all projects.

Thank you