On 10/12/07, Mark Wheeler <[EMAIL PROTECTED]> wrote: > The predomininant recommendations I've seen say allocate two VDISK swap > areas, one with a "best guess" size, and a 2nd lower-priority, large swap > area. Since Linux supports up to eight, I define 'em all, starting with a > small, high-priority VDISK, then doubling each successive disk thereafter,
Sure, you can. From a systems management point of view it is good if you can tell from the outside which VDISK should not be used. I think we came up with the "next disk double size" in one of the residencies over a beer. It looked very scientific... (later than evening we argued that 3.14 would be a good factor too). If you don't know enough about the application requirements, your guess is as good as mine. If you have more information, then arbitrary doubles may not be the optimal. Ideally, what you want is that the sum of virtual machine size plus the used swap disks match certain levels of resource usage by the server. Suppose the server runs two different workloads, one 200M and one of 300M, then it would be nice to have that extra 100M in one VDISK (that will be paged in by z/VM when the 2nd workload is running and can be paged out when it does not). There's also a bunch of pages that Linux can swap out immediately when you push it a bit, and those will not be needed any time soon. If that fills up the first levels of your swap pyramid, then you might as well fold some levels from the 8. Rob -- Rob van der Heij Velocity Software, Inc http://velocitysoftware.com/ ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
