For process models using non-volatile tracking, process and join timer information is stored in the Process Engine database component. If the host Integration Server terminates unexpectedly or is restarted, the timers can be recovered and restarted.

When the timer is created the expiration timestamp is stored; when the timer is recovered, that same expiration timestamp is used. Note that the Integration Server down time is effectively included in the processing time of the timer; if the timer expires while the Integration Server is offline, when the Integration Server starts, the timer will execute the expiration logic as expected.