On Wed, Feb 22, 2012 at 11:31 PM, Damian Gallagher <damian.gallag...@oracle.com> wrote:
> Well, it's important to understand what we mean by " long idle periods on the > database engine " - in the case of Oracle, the majority of the user work is > done by the user process, rather than a centralised engine, and that user > process is typically only active 1-5% of the time in an OLTP system - leaving > plenty of opportunity to manage the PGA. I'm afraid I encouraged you by mentioning the SGA and PGA and my statements were too generic to justify that. I was painting the simplified picture of a workload that needs all memory when active and nothing when idle. You're right that on a finer level the workload consists of different portions of virtual machine memory, each with their own utilization and duty cycle. But as long as something is active in the virtual machine, z/VM memory management will try to leave the virtual machine alone since it's "still working on it." This means that combining different workloads in a single guest make it more expensive. The issue has lots of variables and is too complicated to resolve in few emails on the list (apart from boring many people to death). I think it would be very good if some of us would be able to discuss this over a beer or two. I'm game if we can come up with a venue. Rob ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 ---------------------------------------------------------------------- For more information on Linux on System z, visit http://wiki.linuxvm.org/
