1) use sp_whoisactive to detect right-now activity and metrics

2) use differential wait stats for say 3 minutes to detect waits that occur during that time

3) use differential file IO stalls for say 3 minutes to detect IO stalls that occur during that time

One or more of those will point you to the issue(s) you are having. As other's have said, there are a BAJILLION things that could be at play here. Initial leading candidate is synchronous mirroring (AG). That is easy to observe (and easy to fix if you can improve the bottleneck, which is almost always IO/RAM on secondary or perhaps network latency/throughput if not in same data center).

If you need scripts to do 2 and 3 above reply here. I have a Tune Like a Guru session I have done at a number of SQL Saturday events I can point you to. sp_whoisactive can be downloaded from sqlblog.com. v11.11 is latest AFAIK.