Component Metrics Monitoring

The health of the component can be determined by the percentage of free resources that are available to the component. For each component, the resource definition could be different.
Component Type Resources to monitor
Traffic Manager Available threads in a pool, available heap, etc.
SQL Available connections, available InnoDB buffer pool, available QCache, etc.
Cache Available connections, connection throttling, available memory, Eviction rate, Hit Rate, etc.
NoSQL Active and Pending Tasks, Dropped Messages, Read and Write Latencies, Replication Lags, Available Heap, Available Disk, etc.
Log Available threads to process API requests, available heap size, Garbage Collection time, etc.
If the component metrics are out of normal range, the components can be tuned to perform optimally, and also be scaled horizontally or vertically. Horizontal scaling means adding more instances of components (except for SQL); vertical scaling means adding more resources to existing components to serve optimally.