Why is Max % RAM Committed capping at 80% in Qliksense Operations Monitor Dashboard despite QMC Max Limit being set to 90%?

[sahilp]

Hello Community,

I am currently investigating a server performance and Qliksense application slowness issue, and I have encountered a memory behavior in the Operations Monitor dashboard that I cannot fully explain.

Environment Configuration:

• Total Server RAM: 128 GB
• QMC Memory Settings:
  • Low Memory Limit: 70%
  • High (Max) Memory Limit: 90%

The Observation:
When monitoring the data under the Performance Summary sheet in the Operations Monitor dashboard, I noticed that the Max % RAM Committed metric hits exactly 80% and never goes above it.

To clarify, this 80% cap is relative to the 90% Max limit set in the QMC (meaning it is capping out at roughly 80% of that 90% threshold, not the physical hardware maximum).

My Questions for the Community:

1. Is there an internal Windows OS limitation or an underlying Qlik Sense Engine setting/working set metric that actively throttles or caps memory commitment at 80% within the operations monitor logging?
2. Why would the Qlik Engine not utilize the remaining allowance up to the 90% High Memory Limit, even when users are actively experiencing application slowness?
3. Are there specific log fields or telemetry metrics I should check to see if the engine is triggering a specific memory clearing/garbage collection routine right when it hits this 80% mark?

Any insights, technical documentation links, or troubleshooting steps to help me understand this threshold behavior would be highly appreciated.

[Andrea_Bertazzo]

Qlik Sense does not put a 80% threshold as far as I know. It is possible that this limit is never reached because the bottleneck is somewhere else, for example in the CPU usage. 
I would suggest the solution in this post for a better understanding of memory usage in Qlik Sense.

[Mohsin4]

I’d be careful about treating the 80% figure as a hard RAM limit without checking the underlying Windows and Qlik Engine metrics. Comparing the Operations Monitor value with actual committed memory and the Engine logs at the same time should help clarify whether this is a reporting difference or an actual resource threshold.

I’ve run into similar questions when looking at hardware configurations, including Viper systems and all-in-one machines, where the displayed resource figure doesn't always tell the whole story. In this case, though, I’d focus on the Qlik and Windows counters before changing the QMC settings.