Hi, Performance between xs and kvm is similar, imho it should not be a reason to change, this goes for most other hypervisors. You will just make your setup even more complex and harder to maintain and troubleshoot. Don't do it.
Just make sure in all cases you use paravirtual drivers in your guests and they'll run just fine (it's called virtio for KVM, most OSes support it - linux and the BSDs out of the box, windows requires drivers installed). If I were to migrate to xs I'd do it for the feature set, like VM snapshot and live migration (without the need of shared storage that is); on the other hand you will lose the power of having a proper OS as your hypervisor with all that entails; hence you will also lose the ability to use GFS/CePH/Glusterfs/CLVM/anysharedmountpoint/etc, the same goes for memory deduplication (KSM). My philosophy is pick one hypervisor and stick with it, test the hell out of it until your setup is solid and you can control it. Lucian -- Sent from the Delta quadrant using Borg technology! Nux! www.nux.ro ----- Original Message ----- > From: "Vladimir Melnik" <v.mel...@uplink.ua> > To: users@cloudstack.apache.org > Sent: Wednesday, 29 October, 2014 10:43:50 > Subject: XenServer > Dear colleagues, > > We're using KVM as a hypervisor in our cloud, but now I'm considering to > create one more cluster based on XenServer, because I've heard it has much > better performance than KVM. So, I'd like to ask a couple of questions and > your answers will be really appreciated. > > 1. Does XenServer really shows better performance than KVM? > 2. Are there other benefits that XenServer provides comparing to KVM? > 3. I'm using shared mountpoint to GFS2 (accessible by iSCSI) as a storage, > what is better to use with XenServer? > > Thank you so much. > > V.