matthew at wil.cx
Fri Jun 27 09:29:37 PDT 2008
On Fri, Jun 27, 2008 at 09:26:08AM -0700, Greg KH wrote:
> On Fri, Jun 27, 2008 at 09:50:18AM -0600, Matthew Wilcox wrote:
> > The ability to trace userspace programs is the key one for the postgres
> > people. They get reports saying how long queries took to execute, which
> > tables were accessed, how long locks (postgres locks, that is, not
> > kernel ones) were held, cache hits, cache misses, etc. It's _really_
> > useful for Postgres developers and admins. And they only get it on
> > Solaris and MacOS, not Linux.
> Isn't this what oprofile on userspace programs is for? Or am I totally
> mistaken here?
Sorry, 'cache' means something different to Postgres people; they don't
mean L2 cache, they mean hits on the gigabytes of tables cached in ram
instead of going out to disc. This is all a lot more high level than
oprofile can handle (unless I don't know about some oprofile
Intel are signing my paycheques ... these opinions are still mine
"Bill, look, we understand that you're interested in selling us this
operating system, but compare it to ours. We can't possibly take such
a retrograde step."
More information about the Ksummit-2008-discuss