Home
| Calendar
| Mail Lists
| List Archives
| Desktop SIG
| Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings Linux Cafe | Meeting Notes | Blog | Linux Links | Bling | About BLU |
On Fri, 25 Jul 2008 22:46:24 -0400 Christoph Doerbeck <[hidden email]> wrote: > The requirement for VTx is only if you plan to use full virt (which has > always been the case). I do not believe there is such a case for para > virt. The only 'requirement' is for PAE, which you won't find (for > example) on older pentium mobility processors. Please correct me if I > am wrong on this. > > Regarding upgrading the host and winding up with broken guests... I know > that on Fedora 5 (I think) there was a time when upgrading the host OS's > (Dom0) xen kernel could have left you in a bind because it introduced a > dependency for PAE support in the guest OS's (DomU). Thus if you didn't > upgrade the guests first, you where hosed. > > Otherwise, I've been pretty successful as far as I can recall with > running PV guests and hosts with dissimilar versions (within reason). > But I'd expect that if you want to make use of live migration, suspend, > resume, etc... I'd probably advise thoughtful planning and upgrade > schedules to keep things closer in sync. > > Last, I don't know if everything Red Hat has done filters down to > CentOS ... [but assuming it does] ... Part of what Red Hat did when > integrating Xen into Enterprise Linux, was to insulate the user from the > hypervisor implementation. Meaning, the tooling for managing virtual > machines is built on project libVirt ... and essentially ... down the > road if Xen is not the "in" technology, then it can be replaced by (or > augmented with) kvm, or what ever ... and your tools, scripts, knowledge > stays the same.
BLU is a member of BostonUserGroups | |
We also thank MIT for the use of their facilities. |