CPU utilisation
Primary
Replica
CPU that sits at or above 90% for long stretches means the cluster is out of headroom and should be scaled up. Read the sizing guide
Cache pool allocation
How cache pools work
Vector cache: 50%
Row buffer: 50%
Cluster size
$42
Resizes roll through the cluster live — no downtime, no dropped connections.
Additional replicas
Every cluster ships with 2 replicas; add more to spread read traffic.
Monthly namespace cost
$42
Delete namespace
This namespace is the default for atlas-core and cannot be deleted.