Which memory metric is important for server monitoring? When should alerts be triggered?

The following metrics are important for monitoring memory on a server:

• RAM usage — total amount of used memory.
• Free memory — how much memory is available for new processes.
• Swap usage — usage of the swap file, an indicator of RAM shortage.
• Cache and buffers — memory used for caching, usually not critical.
• Memory pressure — if available, shows the load on memory.

When to start alerting:

• If free memory drops below a critical threshold (e.g., 10-15% of total).
• If swap is actively used — a sign that RAM is insufficient.
• If there is a sharp increase in memory usage without release.

It is important to consider the specifics of the application and server. For example, for databases or JVM applications, internal memory metrics (heap, garbage collection) should also be monitored. Alerts should be configured to minimize false positives, considering normal peak loads.