Hello JC and al.,

will your fixed size prefix patch (per prefix detailed statistics) be
finally merged into mainline ?

Thanks.

Regards,
Didier.


On 24 juil, 11:37, JC <[email protected]> wrote:
> By the way, is this patch going to be integrated in the mainline?
> I mean, I would understand the whole feature to be kicked out but if
> it kept in the official version, the opimized way to do things sounds
> better, no?
>
> cheers,
> Jean-Charles
>
> On Jun 1, 8:57 pm, Trond Norbye <[email protected]> wrote:
>
>
>
> > On Jun 1, 2009, at 6:53 PM, Jean-Charles Redoutey wrote:
>
> > > I kind of feel lonely on my topic ;-)
> > > Anyway, still in the same area, I've made prefix stats collection  
> > > use the same pattern with per thread statistics and global  
> > > aggregation on request. Those stats overhead shall then not be more  
> > > than the other ones.
> > > Which by the way lead me to a question:
> > > Is it really useful to have per thread mutex? I mean, per  
> > > construction, this is only updated from one thread at a time, isn't  
> > > it? The only concurrent access is the global aggregation, in which  
> > > case atomic primitives should work perfectly (also I do agree, there  
> > > is nothing very portable :-( ), or even no locking at all could be a  
> > > solution, I mean aggregation may not see the "right" values because  
> > > of cache(s) and/or reordering stuffs but they would'nt be "so far"  
> > > from the true, which is already what stats command are answering  
> > > considering the thoughput usually going through memcached servers ...
> > > cheers,
>
> > As they say: premature optimization is the root of all evil.. When we  
> > moved the status to a thread-local scope we wanted to make the stats  
> > "just as safe" as the old implementation, and optimize it later on if  
> > it ever turned up as a performance problem in a benchmark....
>
> > Cheers,
>
> > Trond

Reply via email to