That is a complicated way to do things: it would be a lot easier if instead you had two tables:
Job_id
StartTime
EndTime
And
JobId
HoldTime
ResumeTime
And you fill out a single record for a start and end of each hold, and a single job record for the start and end.
You could then work out the cumulative Hold times very easily for each job, and subtract that from the difference between the End and Start times for the job.
That's pretty trivial!
Doing it with your datastructure is a lot more complex - because you need to correlate each hold with it's matching resume, and that gets pretty difficult, unless you introduce a sequence number or similar.