> Stefan Parvu <stefanparv...@yahoo.com> wrote: > > > Interesting. Do you have any comparative numbers > between > > OSOL and others: RHAT, Ubuntu for instance ? > > > > I had the impression SMF did improve things. As I > read your post > > it seems, sometimes in past but not anymore ... why > is that ? > > Probable somebody should fill in some bugs > regarding this ? > > Correct, it _did_ improve things. But in 2004 there > have been aprox. 60 > services and now we have aprox. 240 services. > > I have no idea in whether service creation is slow, > whether the problem is > that more configuration data has been moved into smf > or whether there are > other reasons. > > Jörg
If SMF starts too many services in parallel, wouldn't that actually slow things down, with more context switches due to time slicing? If so, would the answer be to pace starts proportionate to the number of hardware threads available to the zone, or would it work better to throttle based on the rate of involuntary context switches? -- This message posted from opensolaris.org _______________________________________________ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
