I guess I was wondering if you could achieve something similar to what OS/390, z/OS, and z/VM do with paging to expanded storage versus DASD. Obviously expanded storage is faster for paging, but once a page has been unreferenced for a while you would not want to keep it there but move it to dasd. My thinking was that I could let VM take on some of the responsiblity for paging a linux guest, but, if the linux guest's paging/swapping demands exceeded a certain level I would let dasd swapping take over. Obviously performance would start to suffer for the linux guest, but that would be better than running out of swap space completely wouldn't it? Which leads me to another question.
What does linux do when it runs out of swap space? Lonny Sivey -----Original Message----- From: Rob van der Heij [mailto:[EMAIL PROTECTED]] Sent: Wednesday, May 01, 2002 4:00 PM To: [EMAIL PROTECTED] Subject: Re: V-DISK swap space? At 03:15 02-05-02 +0800, John Summerfield wrote: >I suspect it's smart enough to use them in the order you say;-) It says that new blocks are allocated on a device when the devices with higher priority are are exhausted. My feeling is that you would need page migration as well if you want to exploit a small fast swap device. But still, I do not see a reason to allocate dedicated slower disk space for swap when you can have it all (except for the fact that swapping into v-disk is seen by CP and recorded into monitor records to guide you in planning your virtual machine sizes...) Rob
