Go into your vCenter (if you have more than one, you will need to do this for each.)  Go into the “Hosts and Clusters” view.  On the left pane, select the vCenter Server itself.  On the right pane, select the “Virtual Machines” tab.  You can optionally click the “State” field title to sort by state.  You may also click the host field to sort VMs by the host names (I would recommend this if you have multiple clusters with multiple editions of ESXi).  You can then right-click the virtual machine titles and add the field for “Memory Size” as shown below.  Right-click right on the word “State” in the title of the column.

Once the Memory field has been added (it will probably be far on the right), drag the filed so it’s just to the right of the “State” field.  Now go to the “File” Menu at the top of the vSphere client and select “File” then “Export” and then “Export List”.  Export the file selecting “Excel Workbook” as the file type.  Once exported, open the list in Excel.  In Excel, add a column to the right of the memory column like so:

Edit Cell D2 and put in this formula:   =IF(B2=”Powered On”,VALUE(LEFT(C2,(LEN(C2)-3))),0)

Copy this formula down the entire column.  This checks to see if the VM state is powered on (you do not draw from your vRAM license if it is not).  It then removes the “MB” and converts the value to a numeric so you can sum them up.

Scroll down to the last row and edit the cell in the next empty row in the memory column to something like this:  =ROUNDUP(SUM(D2:D65)/1024,0)

Where D65 is actually the last cell with the memory data in it, your row number will vary depending on how many VMs you have.

The ROUNDUP will round up the memory allocation (in case you have some VM’s with 4000MB allocated to them instead of 4096MB) and you need to divide the sum by 1024 to convert to GB of vRAM.

If you would rather run a PowerCLI script to gather the info, you can find a great article on how to do it here.

I hope your number turns out ok.  If it does not, all is not lost.  There are ways to shrink your vRAM footprint so the impact of the licensing is not as bad.  If you would like me to have a look, email me, I can provide a service to see how much vRAM you have allocated but never use.  That may prolong the next license purchase a bit or perhaps soften the expense.  The new licensing does not need to always be negative, maybe we just need to learn how to size our VMs with the licensing in mind.

UPDATE: If you copy and paste these formulas into Excel, the ASCII is different for some reason.  Just backspace over the quotes (“) and readd them.  The quotes are what Excel has an issue with.

Here’s why: When a virtual machine is running, inside the VM, the operating system maps out pages of RAM.  The hypervisor then addresses and stores those pages in physical RAM.  The address in physical RAM typically never matches the address that the VM’s operating system knows.  The hypervisor (specifically the virtual machine monitor or VMM) has to "translate" the fetching and updating of pages of RAM.  This puts an extra "tax" on the CPU and adds to virtualization overhead.  This can specifically be seen in workloads that perform very frequent page table updates.

The solution:  AMD and Intel continue to enhance their CPUs to offload some of the functions that virtualization provides in software today.  AMD and Intel started with their AMD-V and VT Extensions respectively and now continue on to the second generation of virtualization enhancements.  AMD’s is called Rapid Virtualization Indexing (RVI) and Intel calls theirs Extended Page Tables (EPT). What these functions do is manage the mapping of virtual pages to physical pages for fetching and updating memory so that the CPU does not have to do it in software.

Last week, VMware published a very interesting white paper on RVI and how it performs on some of the more demanding workloads (Specifically SQL, XenApp, and Apache).  You can find all the details and the white paper here .