Instances

The Instances page controls how the MQTT broker service scales. From here you pick the machine size, set the autoscaling range, tune the thresholds, and read CPU and memory charts for the service.

Instance settings

The settings are staged and applied through the project deploy flow, so a change here takes effect only after you deploy it. Use the Save button in the page header to stage your edits. The button stays disabled until you change something.

• Machine is the vCPU and RAM tier each instance runs on, for example 1 vCPU / 2GB RAM. The list of tiers is capped by your project plan.
• Minimum Instances and Maximum Instances set the floor and ceiling for how many copies of the broker run. The minimum cannot be larger than the maximum.
• Scale on CPU Utilization is the CPU percentage that triggers autoscaling. When average CPU crosses this value, more instances are added up to the maximum.
• Cooldown for Scaling Up and Cooldown for Scaling Down are wait periods in seconds that the service holds before adding or removing instances again, so it does not react to short spikes.

Monitoring

Below the settings, two charts plot CPU Utilization and Memory Utilization as percentages over time. Pick a time range from 1h to 30d for both charts at once. Use these to judge whether the current machine size and instance range match the load your devices put on the broker.