A discrete metric is a measurement that has both a value and a weight.
Discrete metrics are different from continuous metrics because each measurement is weighted. A discrete metric is analogous to a weighted average and requires that multiple measurements be taken within a single sample interval. For example, LDAP operation response time is a discrete metric, where the actual response time of each operation is averaged, and the number of LDAP operations is provided as the weight. If no LDAP operations occur in a sample interval, the value would be zero and the weight would be zero.
Some continuous and discrete metrics can also report a minimum/maximum value if the measurement is composed of multiple sub-measurements. The minimum/maximum values are aggregated by averaging, so the values reflect the median.
Some discrete metrics can also convey histogram data. Histogram data represents an additional set of measurements that take individual measurements and place them into value ranges. The PingDataMetrics server supports histograms with up to 15 value ranges. Histogram valued samples are unique because they give a picture of the distribution of the values, and because they more precisely answer the question of "How many samples are greater than X?"