Thanks all. I downloaded a trial of Quest Benchmark Factory and started to have a little play. I will carry on playing with this, but i was pleased to find that the profiler windows-authentication limitation i mentioned (as SSCrazy has pointed out) is not valid at all. I was abled to replay my trace and pretty suvccessfully.
The GUI options imply you can run this multi-threaded, and this does indeed seem to work. Maybe i should check the actual number of concurrent, active transactions, however. Certainly it is the closest thing yet to a decent workload!
Taking your points about profiler replay's little issues into account, perhaps i will now combine this trace with Benchmark Factory's features to ensure a multi-threaded replay and also to ramp up the volume further.
Yes! I have long wondered why on many of my servers TPS is the same or more than BPS. But on this server BPS is consistently 8 or more times higher than TPS. This was the same in my trace. A factor perhaps of a application and connectio layer that uses implicit transactions (this is indeed the case)? Also, we have transactional replication in place so i wonder if the replcmds counts as a batch but not as a transaction ... or something like that?! It is actually quite a busy and well-used system (in my opinion!), so i have been curious why our TPS is about a quarter of what I once read was the low threshold for a busy RDBMS (I saw something that suggested a busy database had 100+ TPS).
Anyway, thanks for your replies. I have something to work with now.