And rather than AoE, I should have said NBD, since this isn't SATA. But otherwise I think the idea is an interesting one.
Another option, providing storage virtualization and thin provisioning, aside from LVM would be Ceph: http://ceph.com/docs/master/rbd/rbd/ ...which is designed to work as a remote virtual block device (or file system, or object store) in the first place. On Fri, Dec 6, 2013 at 10:38 PM, Christian Paro <[email protected]>wrote: > Crazy thought... > > ...you could create a Linux LPAR or VM that manages a large LVM pool with > thin-provisioned volumes, and export these volumes as filesystems over NFS > or as block devices with AoE. > > Then you could build your "thin" Linux VM guests with a small boot volume > (possibly even a read-only shared one) and their "/" filesystem mounted > over the NFS or AoE (given that you've configured your kernel/initramfs to > support the chosen protocol). > > The LVM thin snapshot mechanism could even be used on the storage host to > create fast-copied Linux guests with a shared base image that is only > amended in a copy-on-write manner for those portions of the volume which > are changed by that guest as it runs. Given a big memory cache on the > storage host, this could even help provide the benefit of shared in-memory > caching of all the common OS/application binaries included in that base > image. > > And the model from Mike MacIsaac's "Sharing and Maintaining *" papers > could be adapted over this model to provide on the mainframe a lightweight > provisioning experience much like what can be had with container systems > like Docker/CoreOS on distributed - except with the security benefits of > full virtualization under z/VM. > > > On Fri, Dec 6, 2013 at 11:34 AM, David Boyes <[email protected]>wrote: > >> SFS pretty much does exactly that -- for CMS users. You can provide >> access to files stored in SFS for Linux via the CMS NFS server. Not exactly >> high-performance (dispatching 2 or 3 virtual machines to handle each >> transaction is kinda heavyweight), but it works. >> >> Working on something better. 8-) >> >> > I would like z/VM to provide a capability to add up DASD devices into a >> kind >> > of large pool and place image files like qcow2 (or something similar) >> in it. >> > Wether this image is presented as ECKD or something different to the >> virtual >> > machine doesn't really matter to me. I don't know wether this wish is >> > realistic, but i like this feature on my Linux/x86 environment - >> although i am a >> > System z guy for 20 years by now. >> >> ---------------------------------------------------------------------- >> 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 >> ---------------------------------------------------------------------- >> For more information on Linux on System z, visit >> http://wiki.linuxvm.org/ >> > > ---------------------------------------------------------------------- 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 ---------------------------------------------------------------------- For more information on Linux on System z, visit http://wiki.linuxvm.org/
