A cluster tier is an Atlas deployment size that sets the memory, storage, vCPUs, and IOPS available to each data-bearing server. Choose a tier for the application’s current workload, then resize it as demand changes. section 10.0
Match the tier and configuration to the workload:
- M0 suits beginners; Flex suits development and testing. M10 and M20 suit development or low-traffic applications, support replica sets, and can handle occasional traffic increases. M30+ tiers are intended for production. section 10.0 section 10.2.1
- M10+ clusters allow storage-capacity customization. For M40+, choose among General, lower-CPU, and Local NVMe SSD classes according to compute, storage-performance, and cost needs; available options vary by provider. section 10.2.4
- Tier autoscaling can respond to CPU and memory use within configured minimum and maximum sizes. It stays within the same cluster class and does not add replica-set nodes or sharded-cluster shards. section 10.2.3
Resize an M10+ cluster
Use the Atlas CLI command `atlas cluster update <your cluster> --tier <any tier larger than M10>` to change its size. Alternatively, set minimum and maximum sizes under Cluster Tier Scaling in the Atlas UI to let Atlas adjust the tier automatically. Tier autoscaling is enabled by default for clusters created in the UI, but disabled by default when created through the API or CLI. section 10.2.1 section 10.2.3
