DX Unified Infrastructure Management

Hi All,

 What is the best way of monitoring virtual environment  we have a mixed flavor of vm  and physical machines, the virtual machines have robots installed and it is monitored by vmware probe too ,does it cause any impact ? In UMP those severs are dedicated to virtual machines as well as Device ,confirm whether it may lead any issues ?

Regards,

If you monitor a Virtual machine with both a robot and vmware probe at the same time and its the same set of metrics ..you end up collecting similar QoS information. But this normally happens in quite a many environments 

This again comes down to Agent vs Agent-less monitoring discussion and each has it own advantages

But making a choice between vmware and robot depends on how critical the machine is and what kind of metrics you decide to monitor on VM

If its the basic stuff related to cdm its good to go with vmware, but if the machine is bit critical and you decide to have applications running on the server also to be monitored , its good to go with a robot. 

Again vmware probe provides some good high level visibility into your virtual environment , datastore, resource pool etc., 

In case you are a service provider and there are some SLA's tied to the monitored QoS , robot would be a good option to monitor servers because of various advantages uim robots provide 

You can go with a combination of both but try avoiding collection of duplicate metrics at the VM level and if the server is a critical one and needs more attention.. robot would be a better choice . Sometimes I feel the info provided by vmware probe from VM's is unreliable 

I'd recommend that you use Robot for specific VM (as if it was a physical box) as far as you don't have monitoring needs with vSphere Server / Hypervisor aspect.

In my opinion it is very interesting to have both points of view for every virtual server. Some of the performance metrics that you can monitor in VMs, can have significant differences if you monitor them from inside the guest operating system, or if you monitor them from the Hypervisor point of view. This is especially important for CPU, Memory and swap, on the other hand, if you do not monitor VMs with the vmware probe, you are not going to be able to see which VMs are running in which hypervisor in the UMP.

I think that the best approach for virtualized scenarios is to monitor VMs with robots and the vmware probe, and try to correlate them into the same master device, so that all metrics (robot metrics and vmware metrics) are displayed within the same device.  Make sure that "vmware tools" are installed in the VMs, this allows to know the IP when the vmware probe discovers the device, and makes easier to correlate both devices properly.

The impact of this approach is that you are going to store more QoS performance metrics for the same server, so you will need more space in the DB, however you are going to be able to seamless monitor all the virtual servers and offer and unique view of all virtual infrastructure.