An NVMe solid-state drive provides high-speed, low-latency storage for demanding applications that require high-throughput I/O. In Atlas, it is a cluster class available for dedicated clusters on AWS and Azure. section 10.2.4
Choose NVMe storage when the application justifies its operational and pricing trade-offs:
- Use it for demanding workloads needing low-latency, high-throughput I/O. section 10.2.4
- Compare it with General clusters for balanced production resources or Low CPU clusters when memory matters more than CPU. section 10.2.4
- Confirm the deployment region: Atlas supports NVMe clusters on AWS and Azure, but not currently on GCP. section 10.2.4
Atlas storage and oplog planning
For a dedicated M40-or-larger Atlas cluster, select the local NVMe SSD class when I/O latency is a priority. If storage autoscaling is disabled, Atlas allocates 10% of disk size to the oplog for NVMe clusters; with autoscaling enabled, configure the minimum oplog retention window as needed. section 10.2.4
