[Mulgara-dev] Transaction commit performance

[Life is hard, and then you die]

On Fri, Jun 13, 2008 at 05:05:17PM -0400, Alex Hall wrote:
> Life is hard, and then you die wrote:
> [snip]
> >> One thing that occurred to me is that it might benefit us to keep track 
> >> of which byte buffers have been modified in the current transaction, and 
> >> only force those to disk (the current approach is to force ALL mapped 
> >> byte buffers).  On further thought, though, forcing a non-modified 
> >> mapped byte buffer should be a no-op (assuming that the OS is smart 
> >> about managing its cache).  Do you agree?
> > 
> > I had a discussion with Paul about this a while back because I was
> > thinking the same thing, and he rightly pointed out that it's a no-op
> > (I even briefly played with some mods to track and flush only dirty
> > buffers, and it didn't help, at least on Linux).
> 
> Would you by any chance still have those mods available?  I would be 
> interested in seeing if they have any impact on Windows performance.

Sorry, looks like I completely deleted it again - I didn't find
anything in my backups either.


  Cheers,

  Ronald