>>> On 8/18/2015 at 08:43 AM, Jon Butler <[email protected]> wrote: 
> It seems to me this is another abstraction layer that would enable an 
> instance of bare-metal Linux to spool up multiple instances in an LPAR 
> without z/VM.  

That's the whole purpose of KVM, on any architecture.

> What would the effect be of running KVM in a Linux instance under z/VM?

Pretty bad performance, to start with.  The z/Architecture only provides 2 
"layers" of SIE (start interpretive execution).  PR/SM takes up one, z/VM would 
then take up the second, leaving Linux without that huge boost.
   
> And of course PR/SM is still dispatching logical IFLs onto real IFLs...or are 
> we eliminating PR/SM on the LinuxONE and perhaps going back to a version of 
> MFT?

Not at all.

> I will be interested to see the Nürburgring lap times comparing KVM with 
> z/VM.

Things will probably not be too different between them at low levels of over 
commitment of resources.  After that, z/VM will blow the doors off KVM.


Mark Post

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to