Dear folks,

We've uploaded the updated version of our MIB proposal about virtual
machines controlled by a hypervisor here:
http://tools.ietf.org/html/draft-asai-vmm-mib-03

The followings are the known issues related to notifications listed
on P.42 of our I-D.  We would like to discuss the followings (and
other issues you find, of course), so please review the I-D and
post your comments to the list.

o  Issue 1-1) Scalability issue on notifications: The number of
   virtual machines managed by a bunch of hypervisors in a datacenter
   possibly becomes several thousands or more.  If these virtual
   machines frequently change their administrative state, many
   notifications could be trapped.  Since an SNMP manager has to
   handle SNMP traps of these notifications, there exists a
   scalability issue on handling them.  Should we add some
   `vmXXXNotificationEnable' objects to disable traps for each
   notification?  Or any other ideas?

o  Issue 1-2) vmDeleted: Is `vmDeleted' required?  If the virtual
   machine is not persistent on the hypervisor, its entry will
   disappear when it has shutdown. `vmShutdown' can trap the event of
   shutdown of a virtual machine.  So do we remove `vmDeleted' and
   change `vmShutdown' to carry `vmPersistent' in order to
   distinguish ``just shutdown'' and ``shutdown and automatically
   deleted''?

o  Issue 1-3) vmOperState carried with each notification: In our
   current proposal, each notification corresponds to the new
   operational state of a virtual machine, and `vmOperState'
   indicates the old operational state.  For example, when a virtual
   machine is switched on, the operational state is changed to
   running from shutdown.  In this case, vmRunning with shutdown
   vmOperState is be generated when the operational state of a
   virtual machine is about to enter running state.  Is this simple
   and reasonable?


Thank you.
Hirochika Asai

-- 
Hirochika Asai <[email protected]>, The University of Tokyo

_______________________________________________
OPSAWG mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/opsawg

Reply via email to