Holdover is the state that a clock transitions to whenever the input reference clock is lost. It has been shown that providing good, long-duration holdover to a failure of the reference signal is neither easy nor inexpensive. However, there could be multiple reference clocks simultaneously available to a slave clock, offering the ability to provide backup signals in the case of failure. Consequently, network designers prefer a design where multiple references are available to the network elements—providing redundancy and backup for timing synchronization.
The question arises, how should the clock behave when an input reference clock is lost and the clock decides to select the second-best clock source as the input reference? The behavior during this event is characterized as the transient response of a clock.
Multiple reference inputs to a clock could be traceable (via different paths) to the same source of time (such as a PRTC) or back to different sources. The clock quality information (such as Ethernet synchronization messaging channel (ESMC) and PTP packets) sent through the network shows the quality of the source of time that these separate references offer. The software managing the slave clocks then decides which backup reference clock to use based on the quality information. As expected, a high-quality reference is preferred over a lower-quality reference clock.
For cases where there are multiple inputs, the network designer might assign the input references a priority, so that the slave clock can decide in what order the input is preferred. So, when the highest priority (say priority 1) reference clock fails, the slave must decide whether there is another reference available to provide an input signal. The slave selects one of the remaining sources available out of those signaling the highest quality level. If more than one source signal is available at that quality level, the slave selects the available reference with the highest priority (say, priority 2). The slave will then start to acquire this new input signal and continue to provide an output signal but now aligned to the new reference.
This switchover, referred to as reference clock change, is known as a transient event and the change in timing characteristics of a clock during such an event is referred to as transient response. This is a very different event from a clock moving into the holdover state, because the change in reference clock happens very quickly, resulting in the slave clock being in the holdover state for only a very brief time.
These transients are thus divided in the ITU-T recommendations into two categories: long-term transients and short-term transients. Switching between two reference inputs is an example of a short-term transient. Long-term transient is another name for holdover.
Transient response specifies the noise that a slave clock generates during such events. Slave clocks are unlikely to be transparent during such events and could either transfer or generate additional noise, which is measured as the transient response. Again, for each clock type, there are MTIE and TDEV masks in the ITU-T standards to specify the accepted transient response. For example, you can find MTIE and TDEV masks for transient response for clocks supporting SDH (E1) and SONET (T1) in clause 11 of ITU-T G.812.