I have attempted to measure CPU overhead due to inter-zone networking using various tools, including DTrace and an internal network benchmark tool. Any overhead was so small that it could not be measured, and was always less than 1%. This makes sense, because there is no extra IP stack processing if a packet is passing between zones on the same system.

I suspect that the performance improvement that you measured is actually caused by something else. For example, is the system using less memory now? Could it have been paging before, using zones, and is no longer paging?

Or perhaps were you using resource controls and reducing the amount of CPU capacity available to to zone B?

There are many other possible explanations for the performance improvement that you experienced.

Gino Ruopolo wrote:
Hi all,

S10u2, 2 zones. Zone A:  higly optimized Apache+PHP serving about 1000
concurrent users Zone B:  MySQL database

We found that just moving the MySQL database from zone B to zone A reduce
the load of about 20% (system %cpu)!!

Why zones introduce so high overhead??

thanks, Gino

Jeff VICTOR              Sun Microsystems            jeff.victor @ sun.com
OS Ambassador            Sr. Technical Specialist
Solaris 10 Zones FAQ:    http://www.opensolaris.org/os/community/zones/faq
zones-discuss mailing list

Reply via email to