On Sat, 2005-12-03 at 11:49 -0500, Tom Lane wrote: > Simon Riggs <[EMAIL PROTECTED]> writes: > > These changes have almost no negative impact on run time performance and > > can be implemented with minimum change. We can discuss whether the false > > sharing phenomena actually occurs, but the bottom line ISTM is that if > > we can avoid it ever occurring, for almost free, then why not? > > No, you've put the burden of proof in the wrong place. You are > proposing a significant logical complication in the code for a > completely hypothetical improvement --- there is *no* evidence on > the table that cache contention within clog pages is even measurable. > Show us some experimental numbers first.
In a way, I agree with you on the burden of proof. Code wise: I'm not sure this represents a significant logical complication. There would be no more code than there is now, changes would be isolated to about 3 places in two files. There is no evidence either way, is all I would add. But we do have strong indications that it is likely. It's gonna be hard to come up with a smoking gun. We'll have to rethink our performance testing regime to include some larger scale testing with instrumentation. Shelved until measurements indicate requirement. Best Regards, Simon Riggs ---------------------------(end of broadcast)--------------------------- TIP 2: Don't 'kill -9' the postmaster