Re:Comparing Triton containers to VMs and bare metal servers
by jclulow on Thursday June 18, 2015 @12:21PM
(#9739634)
Yes, the QEMU process that underpins the hardware virtual machine lives within a container on the host. It means that a vulnerability in QEMU that allows for guest escape will land you in an otherwise empty container, rather than out on the host where other tenants are accessible.
But Triton is first and foremost about _container-based_ virtualisation: which involves no HVM at all!
Re:Comparing Triton containers to VMs and bare metal servers
by gtirloni on Thursday June 18, 2015 @03:00PM
(#9740872)
Ah, that's something I hadn't considered. Good point. I guess the overhead will be minimum in that scenario.
Re:Comparing Triton containers to VMs and bare metal servers
by xenophonf on Thursday June 18, 2015 @04:30PM
(#9741440)
Is it possible to do the same thing with cgroups or jails?
Re:Comparing Triton containers to VMs and bare metal servers
by bcantrill on Thursday June 18, 2015 @12:19PM
(#9739618)
No, it's not -- a Triton container is a SmartOS container running an LX-branded zone.[1][2]
[1] http://www.slideshare.net/bcantrill/illumos-lx
[2] https://www.joyent.com/blog/triton-docker-and-the-best-of-al...
Edit: I think I misunderstood the question -- apologies!
Can't we just describe VMs as VMs and be done with it? So it's a container that has a VM inside it?