> "Curtis Faith" <curtis@galtair.com> writes:
> > I'm not really worried about doing page-in reads because the
> disks internal
> > buffers should contain most of the blocks surrounding the end
> of the log
> > file. If the successive partial writes exceed a block (which
> they will in
> > heavy use) then most of the time this won't be a problem
> anyway since the
> > disk will gang the full blocks before writing.
>
> You seem to be willing to make quite a large number of assumptions about
> what the disk hardware will do or not do. I trust you're going to test
> your results on a wide range of hardware before claiming they have any
> general validity ...
Tom, I'm actually an empiricist. I trust nothing that I haven't tested or
read the code for myself. I've found too many instances of bugs or poor
implementations in the O/S to believe without testing.
On the other hand, one has to make some assumptions in order to devise
useful tests.
I'm not necessarily expecting that I'll come up with something that will
help everyone all the time. I'm just hoping that I can come up with
something that will help those using modern hardware, most of the time.
Even if it works, this will probably become one of those flags that need to
be tested as part of the performance analysis for any given system. Or
perhaps ideally, I'll come up with a LogDiskTester that simulates log
output and determines the best settings to use for a given disk and O/S,
optimized for size or speed, heavy inserts, etc.
