Dear Mike,

Thank you for sharing the feedbacks from operators.  I think some of them are
already defined in the current version, and some of the rest shall be good to be
defined in the next version.  I will review these objects comparing the current
version of our proposal.

At this moment, I don't understand some of them, so could you please explain 
them?

> Virtual Machine Processor

>          ProcessorUserTime  CPU time spent on user processes (milliseconds)
>          ProcessorSysTime  CPU time spent on system processes (milliseconds)
What's the difference between these two time metrics?
Are they user-time and system-time at a guest OS?


> Virtual Machine Processes
>          PID  Process ID
>          ProcessName  ProcessName
>          ProcessParam  ProcessParameters
>          ProcessCPUUsage  ProcessCPUUsage
>          ProcessMemoryUsed ProcessMemoryUsage
Do you mean they are processes on a guest OS?

If the objects above are those of a guest OS, they need a special tool such as
vmware tools installed to the guest OS.  So, MIB modules on guest OSes, such
as HOST-RESOURCES-MIB, could be used.


The following objects are also related to guest OSes.
> Virtual Machine Partitions
>          PartitionLabel  disk path (c:\, /boot, etc)
>          PartitionCapacity  capacity of the disk in bytes
>          PartitionUsedSpace  capacity minus freespace
>          PartitionUtilization  used as a percentage of capacity

Some of them might be recognizable from hypervisor.  The partition format
itself does not vary by guest OS (i.e., MBR or GPT partition), so some of these
might be able to available from hypervisor.  However, I still don't think VM-MIB
is the appropriate place to contain these objects, and it is better to use MIB
modules on guest OS (e.g., HOST-RESOURCES-MIB).  Moreover,
PartitionLabel, PartitionUsedSpace, and PartitionUtilization are information
on file system, then a hypervisor cannot recognize them without any special
tools such as vmware tools.


Thank you.
Hirochika


On Nov 5, 2013, at 10:02 AM, Michael MacFaden <[email protected]> wrote:

> Hi,
> 
> 
> To add to Jon's request to see 'top' like stats from virtual cpus assigned
> to a given virtual machine, I worked recently finished collecting feedback 
> for a few US fortune 50 enterprises, here is the compiled the following list 
> of 
> metrics they asked for. I think we already have quite a few covered,
> but still think some of these should be included in the next draft...
> 
> Virtual Machine General Metrics
>          MachineName  name of VM
>          MachineServer  ESX server that hosts this VM
>          MachinePowerStatus  power status of the VM
>          MachineUpTime  number of seconds since last boot
>          MachineLastUpdate  number of heartbeats received from the VM
>          MachineGuestOS  fullname of the guest OS
>          MachineGuestIpAddr  IP address of the guest
>          MachineCPUNum  Number of vCPUs in SYSTEM
>          MachineResourcePool  The Resource Pool this VM is a part of
>          MachineLastVMotionTime Shows the last time a VM was vmotion
>          MachineHAEligible  Is this Virtual machine protected using Vmware HA
>          MachineCreatedBy  Who Created this Virtual Machine
>          MachineCreatedDateTime When was this Virtual Machine Created
> 
> Virtual Machine Memory
>          MemoryTotalSize  memory size of the VM in MB
>          MemoryMaxAlloc  amount of memory in MB that the VM can receive
>          MemoryMinAlloc  amount of memory in MB guaranteed to the MB
>          MemoryHostUsage  consumed host memory in MB
>          MemoryGuestUsage  active guest memory in MB
>          MemoryHostUtil  host memory used as a percentage of the total VM 
> memory
>          MemoryGuestUtil  guest memory used as a percentage of the total VM 
> memory
>          MemorySwapToFile  amount of memory being swapped in KB
> 
> Virtual Machine Processor
>          ProcessorIdleTime  CPU time spend in wait state (milliseconds)
>          ProcessorUserTime  CPU time spent on user processes (milliseconds)
>          ProcessorSysTime  CPU time spent on system processes (milliseconds)
>          ProcessorWaitTime  CPU time spent in ready state (milliseconds)
>          ProcessorUtilization  user time as a percentage of total CPU time
>          Processor%Ready  ready time as a percentage of total cpu time
> 
> Virtual Machine Disk
>          DiskDescription  disk label
>          DiskType  type of disk (hard disk, floppy, etc)
>          DiskAccess  read/write
>          DiskCapacity  capacity of the virtual disk
>          DiskRemovable  true/false
> 
> Virtual Machine Partitions
>          PartitionLabel  disk path (c:\, /boot, etc)
>          PartitionCapacity  capacity of the disk in bytes
>          PartitionUsedSpace  capacity minus freespace
>          PartitionUtilization  used as a percentage of capacity
> 
> Virtual Machine Network
>          Description  Network device description
>          PhysAddress  physical MAC address
>          Read Count  number of packets read
>          Write Count  number of packets written
>          Read Rate  number of bytes read
>          Write Rate  number of bytes written
> 
> Virtual Machine Processes
>          PID  Process ID
>          ProcessName  ProcessName
>          ProcessParam  ProcessParameters
>          ProcessCPUUsage  ProcessCPUUsage
>          ProcessMemoryUsed ProcessMemoryUsage
> 
> Next step, break these down from what we have and do not.
> Decide what goes where, not all of course has to go into VMM-MIB,
> for example reporting the process list of a VM might be best reported
> via ENTITY-MIB/HOST-RESOURES-MIB via context.
> 
> Regards,
> Mike MacFaden
> Staff Engineer I R&D Apps, VMware Palo Alto CA
> _______________________________________________
> OPSAWG mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/opsawg
> 

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

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

Reply via email to