Doug,

> I see no good reason for forced writes being off except possibly the odd
> time when lightning fast performance is more important than ensuring
> database integrity. However, I think those times are rare. Given today's
> hardware speeds and capacities does the need to turn off forced writes

ever

> really come up? Does anyone have real, measured data on the difference it
> makes?

I can see in our business that there are some overnight process that may
benefit from this. Currently the process takes about 5 hours, but is
growing, when it approaches 12 hours (6pm -> 6am), we look at the code and
optimise it, normally halving the time whilst doubling the complexity of the
code. One option has always been to stop server, turn forced writes off, do
proc, turn forced writes back on.

I haven't tried with and then without forced writes, but that would be
interesting, maybe I'll give it a go and get back to you, like all tests,
all they ever say is that doing *this* process you can expect *this* speed
benefit.

Good to hear from you, haven't seen many posts from you for a while.

JAC.