This read-only Counter64 reports how much the output discards count increased since the previous stats-polling interval for this interface's live, currently-active error-statistics record, i.e. the per-interval rate rather than the lifetime total. A nonzero value here means new failures of this specific kind are occurring right now on this link, not just accumulated history, since a rising count here on this port almost always means its outbound queue is filling faster than the link can drain, i.e. this specific uplink or interface is oversubscribed or congested. For example, a burst of a few thousand output discards on this port during a nightly backup window is a typical sign that this particular uplink needs more bandwidth or tighter QoS, which this delta field would show as a sudden nonzero reading in a single interval.