YugabyteDB Managed provides the following tools to manage clusters:
- Scaling - To ensure the cluster configuration matches its performance requirements, scale the cluster vertically or horizontally as your requirements change.
- Read replicas - Add read replicas to lower read latencies in regions that are distant from your primary cluster.
- Backups - Configure a regular backup schedule, run manual backups, and review previous backups.
- Maintenance windows - Yugabyte only performs cluster maintenance, including database upgrades, during a weekly maintenance window that you configure.
- PostgreSQL extensions - Extend the functionality of your cluster using PostgreSQL extensions.
Pause, resume, and delete clusters
To reduce costs on unused clusters, you can pause or delete them.
Access Pause/Resume Cluster and Terminate Cluster via the cluster Actions menu, or click the three dots icon for the cluster on the Clusters page.
Deleting a cluster deletes all of its data, including backups.
Paused clusters are not billed for instance vCPU capacity. Disk and backup storage are charged at the standard rate (refer to Cluster costs). You can't change the configuration, or read and write data to a paused cluster. Alerts and backups are also stopped. Existing backups remain until they expire. You can't pause a Sandbox cluster. Yugabyte notifies you when a cluster is paused for 30 days.
Note that if another locking operation is in progress, you can't pause or delete a cluster until the other operation completes.
The following operations lock the cluster and only one can happen at the same time:
- backup and restore
- pause and resume
- scaling the cluster, including adding and removing nodes, increasing disk size, and changing IOPS
- create, delete, and edit of read replicas
- any scheduled maintenance, including database upgrades, certificate rotations, and cluster maintenance (a backup is run automatically before a database upgrade)
- configure metrics export on the cluster
In addition, on AWS, any disk modification (size or IOPS) blocks further disk modifications for six hours (this includes a scaling operation that increases the number of vCPUs, as this also increases disk size).
Make sure that you schedule maintenance and backups so that they do not conflict.