What is the memory threshold percentage that should trigger an incident?

The memory threshold percentage that triggers an incident is set at 25% available memory. This benchmark is established as a best practice to ensure optimal performance and reliability of the system. When available memory falls below this level, it can indicate that the server is under stress and may impact the efficiency of queries and operations, leading to slower performance or even system instability.

Setting the threshold at 25% allows for a buffer that provides administrators with ample time to respond to potential issues before they escalate. It reflects a cautious but proactive approach to resource management, ensuring that there is sufficient memory available for processes and preventing situations where the system could be at risk of performance degradation.

Choosing a higher threshold, such as 30% or 15%, would not effectively signal when action is needed and could lead to delays in addressing memory-related issues. In contrast, a threshold of 10% would be too low and might result in frequent incidents, making it challenging to maintain system performance effectively. Thus, the 25% threshold strikes a balance between being alert to potential problems while still allowing for routine operations to occur efficiently.