The one-way Frame Delay for an egress Service Frame at a given UNI in the EVC is defined as:

"the time elapsed from the transmission of the first bit of the ingress Service Frame at the ingress UNI until the reception of the last bit of the Service Frame at the egress UNI"

This delay definition is illustrated in the figure at right.

This definition of Frame Delay for a Service Frame includes the delays encountered as a result of transmission across the ingress and egress UNIs as well as that introduced by the CEN.

When the UNIs are not time-synchronized, measuring one-way frame delay is not possible. Therefore, an approximation is to take 1/2 of the two-way delay between the 2 UNIs.

Frame Delay Attributes

MEF 10.2 defines three performance attributes for One-Way Frame Delay:

  • One-way Frame Delay Performance corresponding to a percentile of the distribution
  • One-way Mean Frame Delay
  • One-way Frame Delay Range

The Performance objectives are given for a time interval T (e.g. 15 minutes). The Frame Delay measurements are done much more frequently (e.g. 1 per second) This yields several (900 in this example) data points per time interval T . These data points can be used to derive a distribution or some statistical function. 

  • Percentile: if the xth percentile (e.g 95th ) of the distribution over T
  • Mean: the arithmetic mean over all samples for time interval T
  • Range: The difference between two pre-defined percentiles (e.g. 90th percentile to 20th percentile).

Frame Delay Measurement

Frame Delay is measured using the ITU-T Y.1731 DMM/DMR messages for round trip (2-way) delay measurement (single-ended delay). If the network elements containing the two UNIs are time synchronized then the 1DM message for 1-way delay measurement (dual-ended delay) can be used. These are described in Measuring Frame Delay.


Assume an E-Line service with single CoS ID.

The SLS specified for Frame Delay is that for any given 1-hour the 95th percentile will be 50 msec.

The Service Provider measures the (round trip) Frame Delay every 1 second between 2 MEPs in the EVC ME.

Every sample is divided by 2 to obtain the one-way Frame Delay. For each time interval T = 1 hour, the SP computes the 95th percentile. If the 95th percentile is no greater than 50 msec, the SLS is met. Note that this is just for a single ordered pair of MEPs. If the service has multiple pairs, then the Frame Delay would be the maximum as described in the introduction.

Scope of Frame Delay Measurement

Refer to MEF10.2

The reader can refer to section 6.9.2 of MEF 10.2 for formal definitions of these performance attributes.


Related and Further Reading




  1. Anonymous

    Can someone explain why FD, MFD and IFDV are defined in MEF 10.2 in terms of Service Frames, although the PM measurements are based on DMM/DMR messages?  I assume that the definitions are the theory and the measurements are estimates of the theoretical metrics.

    1. Anonymous

      DMM/DMR messages travel the same path of service frames. So, the frame delay an vary only with respect to device processing. Device processing delay is subtracted from the total delay when we use DMM/DMR method. Hence, it can be considered as the delay of a service frame.