Show Table of Contents
16.3. Potential Impact on Server Performance
Theoretically, there is no limit to the number of metrics that can collected or the number of alerts that can be fired.
In reality, there are natural constraints within the IT environment that limit both monitoring and alert settings:
- Database performance, which is the primary factor in most environments
- Network bandwidth
There are no hard limits on JBoss ON's alerting and monitoring configuration since it depends on the number of resources, number of metrics, collection frequency, and the number of alerts.
As a rule of thumb, there are these performance thresholds:
- Up to 30,000 metrics can be collected per minute
- Up to 100,000 alerts can be fired per day (roughly 70 per minute)
Plan how to implement metrics collection and alerting. Prioritize resources and then the information required from those resources when enabling metrics schedules and setting collection frequencies. Then, based on those priorities, plan what alerts are required.
Clear monitoring and alerting strategies can help maintain performance while still gathering critical information.
Additional storage nodes can be added to the storage cluster to extend the amount of metrics collected.
The storage nodes can be permanent, as part of the JBoss ON design, or can be created dynamically, using an alert condition on the existing nodes to trigger a new node deployment.
Creating additional nodes is covered in Section 24.3, “Deploying and Managing Storage Nodes”.