vRealize Operations Manager - Deep Dive Part 3

vROps 6 Preparation and Architecture continued...

I know I said this part would be more installation from part 2 but after experiencing more questions from clients and needing more clarity for the sizing and design I decided to proceed with more design.

Questions about the design

I have vCenter alerting and performance charts within my infrastructure, why would I need something like this?
vROps provides so much more than just alerting and realtime performance charts.   It provides capacity management, recommendations and analysis in clear charts, oversized and wasted resources, scheduled reporting and more.  Is it right for all environments?  Not all, as cost for some small customers may be prohibitive and if the environment is very static and doesn't grow past a couple redundant Virtual Hosts, then monitoring it may seem overkill.

vROps is only for examining virtual environments, right?
No, it is not for only examining virtual environments.  With the inclusion of Log Insight, and Hyperic physical computers, and devices can also be examined within this monitoring solution.

Do I need vSphere 6 to run vROps 6?
No, minimum requirements for vSphere is
  • VMware vCenter Server 4.0 Update 2 and later, managing hosts running ESX/ESXi 4.0 and later
Really?  4.0?  yep, vROps connects to the vCenter server through port 80, 443. I am sure most people are running at least this version, but if you are one of those running vCenter 2.5 and ESX server 3.x, sorry you need to upgrade.

I want high availability on my vROps servers?
This is bit of a loaded question as there is numerous ways to make any solution Highly Available. The biggest question to ask yourself is what is the purpose of making this tool highly available. There should be a business driver to warrant the highly available design consideration.  As by typical setups of vSphere, there is vSphere HA that keeps the systems up and running through sensing the host failure and powering up the VM upon a different ESXi server.  There is a limited downtime but the system brings the node up quickly.  

vROps also offers an application HA that is built into the product by a single checkmark box and the installation of the Replica Node.  At writing the current cluster size limitation is 8 nodes, including the master, replica and data nodes (Remote Collectors is not included in that count.) and with the inclusion of Application HA the node count effectively halves.  

How big does the dataset get and can I run this on slow storage?
That will be answered as we go through the sizing spreadsheet, however notice as you size the structure that IOPS is also calculated.  Inputs / Outputs per second is a calculation for disk speed or how quickly the data will be processed.

Can I resize the data disk attached to the node?
Yes and the process is very simple.  See the article here

Sizing and cluster size

There is a fantastic sizing calculator on the knowledge base article placed in Excel.  Go to KB 2109312 and locate the section labelled Attachments, and download the xlsx file.  There is three tabs in the spreadsheet, Overall Scaling, Sizing guide (basic) and sizing guide (advanced).

The overall scaling shows the same table from the KB article and shows maximums for single node objects and metrics (explained later) and capabilities of multi-node configurations.  As shown the small node handles 2,000 objects or 700,000 metrics per node.  Up to the large node sizing where it can handle 10,000 objects and 3.5 Million metrics per node.  This table only shows Compute and Memory requirements and does NOT show the disk requirements.  Delving into the sizing guide allows further calculations for your storage based upon common parameters.

Basic Sizing Guide

The basic sizing guide sizes based upon a very simple environment.  vCenter Management pack only and an understanding that this will only monitor within your virtual environment. Enter the amount of virtual machines, hosts and datastores to get a summed amount of objects to be monitored.  Likely scenarios would include a monitoring solution only for the VMware / virtualization administrators to ensure the environment is up and running and fix any problems that might occur.  I have seen most environments that use this sizing calculator, end up with a small node and a very straightforward installation.

Advanced Sizing Guide

Even with this sizing guide there is situations that these still don't size well enough.  However with that said this captures 95% of the sizing requirements that the product is installed with.
This sizing delves into how many vCenters, and other vSphere objects, it looks at Horizon (VDI), Hyperic (Application monitoring within virtual machines and physical servers, including Hyper-V machines) NSX devices (software defined networks), storage arrays, and even connections to AWS.


What is an Object?  An object is a vCenter Server, or a datacenter, or a virtual machine.  It is something that contains an item that is relatable. vCenter server is a parent object to a Datastore, and a virtual machine is a child object to ESXi Host.  Is the percentage of CPU used by the virtual machine an Object?  That would be a metric of the object and not an object itself.  So if you had a vCenter server, with Datacenter, three hosts with 20 vms, all contained in three datastores, you would have 29 objects.


A metric is a calculation within an object, such as the example above % of CPU used in a given time. It is something that is recorded and based on a time frame.  So when you are looking at the sizing chart discussed above, take note of the size of your environment and the how much performance you need going forward.

Download the sizing guide and play around with some numbers and get a feel for the size of your environment and what you are all wanting to collect.  This guide will continue with more design considerations in part 4.