There is a huge problem on docker userspace virtualizaion, isolation is not possible and the underlying host is exposed on many ways, it would be perfect to have simple paravirtiualization with dockerfile for similar to creating docker images

Is there a timetable for vTPM 2.0 on DomU?

I ask because the web page at https://wiki.xenproject.org/wiki/Virtual_Trusted_Platform_Module_(vTPM) asserts that Xen 4.6 supports vTPM 2.0; but that page in turn refers to a document at http://xenbits.xen.org/docs/unstable/misc/vtpmmgr.txt that asserts that DomU will continue to run TPM 1.2.

Is that correct? In other words, do the virtual machines continue to run TPM 1.2?

Thank you.

True GPU virtualization is becoming more mainstream. Passthrough is only useful in certain applications and vGPU support would greatly expand the possibilities and reach of Xen.

Make your videos Linux friendly, It's sad I have to even post this... ADOBE, shockwave flash... for real... whatever...

All the examples for installing and using Xen are currently antiquated.
Can someone just record how they set up a new system using Ubuntu 14/15, Xen 4.5.1, and XAPI/xl? Sections on setting up networking, passthru, BIOS bugs and *NIX/Win/OSX would be exactly what is needed.

This is useful for backup devices, transferring data between VMs, etc.

Add a mechanism to remotely replicate the vm's without the use of external storage. This is a major need for smaller deployments.

Support for VM live migration even if some SR-IOV VF virtual devices are attached to VM. such has http://www.youtube.com/watch?v=17so6019Dow

Presently dom0 can balloon down, but does not balloon up if memory becomes available again! dom0 could try to keep the memory at a recommended level by itself.

Note that this feature is already available in XCP 1.6: the request here is to implement this functionality in xl.

I'm using Spice on my xen test system since end of 2011.
Spice is better than vnc for use on maintenance tasks and better than
nx/x2go/rdp installed on domU (even better when used with vdagent and usb
redirection).

I know univention provide an older version of signed GPLPV drivers for Windows guests, but it would be good if there were newer ones available officially off the xen site that were kept more up to date.

When you pass through a GPU to a Windows guest, Windows recognises the device as one that can be ejected from the safely remove hardware feature. It'd be great to allow this to be overriden for some devices.

step 1) Only passthu NVIDIA CUDA (parallel computing) functions.
stap 2) Using CUDA on host-GPU for guest-CPU virtualization on appropriate operations.

... like in VirtualBox or VMware Workstation

In the current architecture, the computing capacity is confined to the single node of the pool servers.
Hasn't been evaluated to introduce a solution for distributed computing?
like Grid Computing or High Performance Computing (HPC)?

It would be nice if we can mount a shared folder, viewed like a device, like in VirtualBox.

For example, to share a folder called "myshare", we run "modprobe vboxsf", then we can mount like "mount -t vboxsf /dev/myshare /mnt".