Redpanda Cloud supports the deployment of Redpanda clusters in single or multiple availability zones (AZs), spanning at most three AZs. Brokers are evenly distributed across AZs, and default_topic_replications is set to 3 by default. Data is evenly distributed across AZs automatically. This behavior is known as rack awareness.

To prevent downtime during cluster upgrades, the Redpanda Cloud cluster operator upgrades one broker at a time. It waits for the health of the cluster to return to its nominal state before continuing with the next broker upgrade, until all brokers are fully rolled out.

Redpanda’s Support, Security, and Site Reliability Engineering (SRE) teams monitor Redpanda Cloud clusters 24/7 to ensure they meet availability service level agreements (SLAs). If incidents occur, teams at Redpanda trigger an incident response process to quickly mitigate them.