This counter reports the total bytes of ingress traffic on this PM instance that were remarked -- allowed through but flagged as exceeding the committed rate while still within the excess rate -- during everything accumulated since the last manual clear. Because this is a life-to-date total that only resets when an operator issues a manual clear (its uncleared TotalStats-bin counterpart never resets at all), a share of yellow traffic that keeps climbing relative to the accepted-traffic total over the tracked span points to a service that has structurally outgrown its committed rate, rather than to one isolated past burst. For example, if this life-to-date yellow-byte total is already a sizeable fraction of the life-to-date accepted-byte total, that's a solid argument for raising the service's committed rate rather than continuing to rely on the excess-rate allowance.