Y.1731 delay measurement implementation ensures the most accurate results under all circumstances. The implementation ensures that there is minimal delay measurement error between packet generation and packet play-out over the Ethernet link.
In order to isolate delay measurement results from the effects of any queuing, scheduling, and shaping procedures, timestamping of source one-way and two-way delay measurement (1DM and DMM) frames in the transmit direction occurs when the first byte of the DM frame is put on the wire (that is, once the actual serialization starts). Last-minute timestamping ensures that DM tests truly measure the delay between two SAP, spoke SDP, mesh SDP, or port endpoints, and not the delay imposed by the routers. Using these accurate measurements, a network operator can separate the delay introduced by the routers from the transmission delay introduced by the transmission network, such as a Metro Ethernet network (MEN) or Generic Framing Procedure (GFP) over SONET links.
Timestamping of received 1DM and DMM frames is similar to transmitted source 1DM and DMM frames. When a Y.1731 delay measurement frame is identified internally, it is immediately timestamped to ensure reporting of only the network-induced delay and jitter.
With two-way delay measurement, delay measurement reply (DMR) timestamping occurs at the entry point to the egress datapath. DMR frames are then classified according to their configured dot1p setting and experience any scheduling delays experienced by the queue and scheduling priorities that the frames are associated with. Performing DMR timestamping at the entry point to the egress datapath provides the most accurate end-to-end measurement of delay and jitter, as it takes into account the internal delay of the responding node due to other higher-priority packets.
In summary, one-way delay measurement performs timestamping in the transmit direction only when the frame is about to be sent on the wire and in the receive direction when the frame is received, before any queuing at the far end. This provides true transport network-induced delay and jitter measurements. Two-way delay measurement takes into account the delay introduced by queuing and scheduling of the far-end node for true end-to-end measurements.