Home > News & Events, Virtualization > How to Save With New vSphere 5 Licensing Model

How to Save With New vSphere 5 Licensing Model

Disclaimer: These are my personal thoughts, and strictly mine.

I missed the big launch of vSphere 5 on July 12 because I was having my vacation. When I came back, I found so many discussions around the vSphere 5 licensing change. It’s understandable that people don’t like changes, especially if the changes may have financial impacts.

Time to learn how to "Google" and manage your VMware and clouds in a fast and secure


Technically, the vRAM pooling simplifies the licensing model, as pointed out by Carter Shanklin. Money wise, some 10-% of existing customers are affected according to this official blog. I think the complaint is more about money than anything else. Since I am not a business person, I won’t elaborate more here.

Following the money talk, let’s look at it in a different way: how to save money with the new licensing model?

In order to answer the question, we need to understand how the vRAM is calculated. Both Al Renouf and Luc Dekens released PowerShell scripts to calculate it. I just pick up the line 23 of Luc’s script here:

$vmRAMUsed = ($VMs | where {$_.PowerState -eq “PoweredOn”} | Measure-Object -Property MemoryMB -Sum).Sum/1KB

The total vRAM is simply a summary of all the $vmRAMUsed of all virtual machines. To save on license cost, you want to drive down the total value.

Based on the formula on how it’s calculated, you have these ways to save money:

  1. Reduce the memory usage of individual virtual machines if possible. To be sure, you want to first check the memory usage to identify these with over allocated memory. Again, check out Luc’s script which lists the stats for each VM. Once you find them, just call ReconfigVM_Task() method to change the memory allocation. Unlike its physical counterpart, a virtual machine’s memory can be increased easily. So you can always add more whenever needed later.
  2. Reduce the number of running virtual machines. You can achieve these in two ways: power off or suspend your virtual machines not in use; consolidate applications from multiple virtual machines to one. In the first case, the vRAM used turns zero. In the latter case, the memory overhead of operation systems is effectively reduced. Note that although you can leverage memory de-duplication and save physical memory, the vRAM usage does not change.

Hope these techniques help you to save with the new licensing model. Please remember, you want to save money, but to the extent that hurts your operations. It’s trade-off you have to carefully balance by yourself.

  1. July 25th, 2011 at 00:07 | #1

    Totally agree. New licensing model will now put financial incentives behind what should ahve already been best practice. Do not maintain VMs that you do not need (VM sprawl), and do not give memory to a VM just because you can. Instead of caving when someone asks for more memory in a VM, actually work with them to determine if it is necessary and educate them on how memory usage in a VM does not need to match what memory usage used to be in a physical environment, you can often do more with less there.

  2. Bjoern Roth
    July 25th, 2011 at 02:12 | #2

    I also totally agree with both of you that VMs shouldn´t get more RAM then needed. But that doesnt make the new Licensing Model better. In almost every configuration I know about not the CPU is the bottleneck but RAM defenitely is. If we hadn´t moved to Enterprise+ on our Site last Year we could not use 256 Gb of our physical available RAM. Thats nothing anyone would cheer about, even with rightsized VMs and even when you know that your not using the 256 GB at all.


  3. July 25th, 2011 at 05:21 | #3

    Beside the so call do not over provision, you may wanna consider some mission critical environment you couldn’t effort to under provision or risk for try and error. Memory over commit and ballooning ensure the physical RAM are always utilize at the optimize level to prevent over provision issue in the pass. Typically for those environment which users cannot effort down time, additional 10 to 20 per cent additional resource provision especially on DISK and Memory will be in place. Well, with new restriction apply to vSphere 5, this will be major impact to those users. Either they need to pay more money or they need to risk them self to down size the machine.

    I think most users understand and well control about over provision issue while they adopted to virtualization. You can claim this is 1 of the way to save cost but it doesn’t make the new licensing looks good to the users.

    You mentioned it only impacted around 10% of the existing customer. I really want to clarify with you on this, since is an interesting figure. The 10 % of customers is base on installed base volume or 10 percent of the client list that VMware have? this can be huge different.

  4. July 25th, 2011 at 06:08 | #4

    Just to be 100% clear that vRAM is actually pooled per license type so you would need to run the code above for each host which has the same license type to get the used vRAM for that license type. My script first checks which hosts are using which license and then works out how much vRAM is being used and how much would be used if they were all powered on.

  5. July 26th, 2011 at 00:46 | #5

    Well said Mark!

  6. July 26th, 2011 at 00:49 | #6

    Thanks for sharing your case Bjoern! The fact of RAM being bottleneck makes it a good candidate to measure real usage of virtual infrastructure.

  7. July 26th, 2011 at 00:52 | #7

    Hi Craig, thanks for sharing your thoughts. For the question, I think it’s great. You may want to post it in the comments of the official blog.

  8. July 26th, 2011 at 00:55 | #8

    Thanks for the clarification Alan!


  1. July 30th, 2011 at 19:45 | #1
  2. September 5th, 2012 at 01:43 | #2