Alexander Skwar wrote:
Hi.
On Tue, Mar 24, 2009 at 11:50, Pradhap Devarajan <prad...@gmail.com> wrote:
Hi Alexander,
Check out the Zones faq
http://opensolaris.org/os/community/zones/faq/#rm_cpupower .
IOW: What I supposed was correct. :)
Was my reasoning also correct
Yes it was. Since you are not having to emulate a complete virtual
platform
and hosting a complete operating system, the expected performance should be
very close to bare metal speeds. After all, it is just encapsulation
of user
space processes and entities (using your terms - I use isolation - but we're
both talking about the same thing). It is not totally free as there are
redundant processes and services. Memory footprint can be slightly
higher, offset somewhat by shared pages when using sparse root zones.
The additional cycle times can be somewhat offset by each instance not
having to do quite as much (20 idle sendmail listeners consume the same
cpu time as 1 idle sendmail listener). You also get much more flexible
resource fencing and control with zones, which is what I particularly
like about them - simple to set up pools and balance load across them.
I know that a colleague is preparing a performance comparison report
and I'm sur
when he publishes it will get mentioned in this forum. At this point I
would say
that you are on the right track with your understanding.
Bob
_______________________________________________
zones-discuss mailing list
zones-discuss@opensolaris.org
