VMware Adjusts vSphere 5.0 Licensing Model
Just a few weeks after introducing its new licensing scheme for vSphere 5.0, VMware has made some adjustments based on customer feedback. These changes have been made to address some of the concerns customers expressed about the new model, which shifted licensing constraints from processor sockets to vRAM (provisioned memory). The feedback that VMware heard boiled down to three key themes:
•The vRAM entitlements may be too low in some circumstances; •The flexibility of the model may introduce a degree of planning uncertainty in some cases; •The change was seen by some as an ‘indirect’ price rise
To its credit, VMware has acted quickly to address the feedback, and it has made changes in three areas. Let’s take a closer look at these changes, why they were made, and what it means to you.
VMware has announced three changes to the licensing scheme for vSphere 5.0:
1.Increased vRAM entitlements on all versions of vSphere 5.0 2.Introduced a per-VM memory entitlement cap 3.Initiated a rolling average vRAM entitlement instead of peak usage to calculate licensing prices Following are the details for each of these changes.
Increased vRAM Entitlements
The vRAM entitlements on all versions of vSphere have been lifted as follows:
•Standard Edition is now 32GB (was 24GB before) •Enterprise Edition is now 64GB (was 32GB before) •Enterprise Plus is now 96GB (was 48GB before) The Essentials and Essentials Plus variants each had their vRAM entitlements increased from 24GB to 32GB per socket, taking the maximum total vRAM entitlements for three 2-socket nodes to 192GB (up from 144GB).
The VMware revised licensing appears to be aimed at addressing the concern that the entitlements were generally set too low. The original entitlement values were based on VMware’s internal analysis of the memory configurations in around 800 installations. Although the company anticipated that a small percentage of users would be negatively affected by the new thresholds, it appears that far more users than anticipated were actually impacted.
What might have tipped VMware’s hand, though, was feedback from customers about what they were planning to do in the future, rather than what they are doing today. This feedback may have revealed that customers were planning to make better use of smaller but more powerful servers with greater memory density than current entry systems. The adjustment to increased memory entitlements is most likely aimed at addressing that situation.
Per VM Memory Entitlement Cap
With vSphere 5.0, VMware introduced the ability to support what it calls “Monster VMs” with up to 1TB of memory. But when combined with the new licensing policy, the increased size of VMs created some practical implementation issues. For example, a 4-socket server running vSphere 5.0 Enterprise Plus Edition has a 384GB (4 x 96GB) entitlement. In short, a user couldn’t run a 1TB Monster VM on a 4-socket server for licensing reasons, unless they had other vSphere licenses on other systems with significant spare pooled vRAM capacity to share. Otherwise, they would need to purchase additional licenses to cover the extra vRAM required.
In response, VMware has put a cap of 96GB per VM for licensing purposes. This means that users can provision more than 96GB on one or more virtual machines, but only the first 96GB of each will be counted for licensing purposes. Using the above configuration example, that means a user could run a single VM with 384GB of vRAM and have 288GB of vRAM capacity to spare, which is a big turnaround from the original situation.
A key objective that VMware has for vSphere 5.0 is enabling customers to virtualize larger business critical applications, and these applications may need large amounts of vRAM provisioned. However, it appears that the original vSphere 5.0 licensing scheme was seen by some as an inhibitor to large, or monster, VM adoption. The increased cap is obviously designed to remove that roadblock.
Rolling Average vRAM Entitlement
The third, and in some ways perhaps the most important change is that VMware will assess the vRAM entitlements based on a 365-day average, rather than point-in-time peak usage. This means that users can temporarily increase memory configurations above their licensed limits, as long as the 365-day average is within the limits.
This capability was of particular concern for test-and-development environments, where it is hard to accurately forecast spikes in usage. In these environments, it didn’t make a lot of sense to provision for rare and out of the ordinary peaks. The switch to average-based entitlements addresses these concerns, and allows companies to select the license that best reflects typical, rather than peak usage pattern, and so gives customers more certainty around planning license fees.
IDEAS Bottom Line
IDEAS believes that the changes made by VMware are pragmatic and sensible. They address what appear to have been the key concerns expressed by users, while still allowing the company to scale revenue with the value that its platform delivers. The move to a 365-day average is particularly interesting, because it introduces a cloud-like pricing model that handles workload peaks and troughs in a way that is more akin to ‘you pay for what you use’, rather than provisioning and paying for the peak usage, which requires a degree of waste to be built-in.
IDEAS thinks that most people would agree that the previous pricing model in vSphere 4.1, which was based on processor slots in a server’s backplane, wasn’t a sustainable business model for VMware as processors with growing numbers of cores allowed more and more VMs to be deployed on individual servers. That model delivered significantly increased performance, and hence value, without an increase in cost for the user. The trick for VMware has been to adjust its licensing model in a way that balances its own needs with the expectations of its customers. These latest revisions will bring VMware closer to accomplishing that goal.
By Gary Burgess
Gary Burgess is SVP Research & Operations at Ideas International and specialises in IT hardware research, including features, performance, Green IT, and pricing as well as IT Infrastructure Support Services research.
Ideas International Limited (IDEAS, IDE:ASX) is an analyst company that provides enterprise IT research, insight, analysis, and tools to both the buy and sell sides of the industry, counting as clients many large technology vendors and major blue-chip global IT users. More information at www.ideasinternational.com