Within a compliance period, a given number of data samples are expected for each of the underlying Quality of Services.
In order to ensure that the compliance calculation is based on a fair amount of data, a concept called Data-Loss Threshold (DLT) is introduced.
DLT is a means of invalidating the calculation based on the actual amount of data available for the calculation.
The Data-Loss Threshold is the percentage of the number of samples that can be missing before invalidating the constrained QoS, thus returning 0% compliance. If the DLT is unused then the compliance calculation is based on all available samples.
E.g. Consider a compliance period of 1 week, and a sampling of disk usage every 5 minutes. This will result in 2016 samples within a period of 1 week (7*24*60/5). If all of the samples are present, then this will fulfill a DLT of 100%. Hence, 1978 samples will fulfill a DLT of 98.11%.