Linode's Xen beta was not as successful as it might have been - I am a participant and the uptimes haven't been that good - although they are improving.
Before it could be deployed in production, Linode needs proper control over disk I/O - more than what just CFQ/ionice provides - to prevent one rogue Xenode screwing the host with swap thrashing or other rogue I/O behaviour.
I personally think that Xen was oversold and has failed to meet the expectations generated by the hype.
caker didn't sound super keen about Xen in the
podcast so don't wait for Xen at Linode.
My guess: UML with SMP and VT/Pacifica support added 'pretty soon now (TM)' will rule the roost at Linode for the foreseeable future.
You heard it here first: 'Netcraft confirms Xen is dying'.
