This page describes operational scaling for SurrealDB Cloud: changing instance size, adding read capacity, and using observability data to time changes safely.
Resizing instances
Resize when monitoring shows a steady mismatch between provisioned resources and workload—for example, sustained high CPU or memory after optimising queries. In Surrealist or the Cloud dashboard, open the instance, choose a new instance type or capacity tier, and apply the change. Expect a short period of elevated latency or reconnects while the platform adjusts capacity; plan changes outside peak traffic when you can.
Adding replicas
Where your plan supports read replicas or clustered layouts, adding replicas improves read throughput and can improve availability. Each replica adds cost; balance redundancy and performance against budget. Confirm replication lag stays within application tolerances after scaling out.
Monitoring before and after
Use metrics (CPU, memory, storage, connections, query latency) and logs to confirm whether scaling solved the bottleneck or merely moved it (for example, from CPU to disk). If utilisation remains low after a large resize, consider stepping down to avoid overspending.
Together with backup and network settings, scaling is part of routine operations—see Monitoring overview for related dashboards and signals.