For those that may have missed it, VMware announced an update to its per-CPU “pricing model”, i.e. licence metric, yesterday. Significantly, it was the virtualisation giant’s fiscal year end on Friday, so they clearly want to get the new year off to a flying start. As is typical of software vendors, VMware couched the announcement in terms which, on the face of it, would suggest that they have their customers best interests at heart. But I think we all know that’s probably not true.
VMware: “Today we announced an important update to our per-CPU pricing model, reflecting our commitment to continue meeting our customers’ needs in an evolving industry landscape. This new pricing model will give our customers greater choice and allow us to better serve them.”
Translation: “We want to make more revenue out of the trend towards higher spec’d multi core CPU chips.”
In truth, this could actually just be the first step on a transition to a core licensing metric for the vendor’s key product sets. VMware are one of the few vendors that still sell products in a CPU metric that only ever means socket CPU chips. Of the software mega-vendors, IBM and Oracle both made the switch to processor cores years ago (although the latter still measures by socket CPUs on Standard edition products using their Processor metric), and Microsoft removed their final processor metric with the transition of Windows Server and System Centre to cores in 2016. This move by VMware shouldn’t be that much of a surprise to industry watchers. So what is the actual change they are making?
The updated metric goes live on 2nd April this year, however VMware are allowing a grace period for any customers in the middle of purchasing hardware for a vSphere platform up to 30th April. After that date, a single licence for any VMware product using the CPU metric, will only cover up to 32 cores on that CPU. Put simply, for each 32 core block on a single CPU chip, you’ll need an additional licence. Considering that, even now, both Intel and AMD have several processors in the marketplace with greater than 32 cores, the costs could soon stack up for enterprises who choose to maintain an on-premise datacentre presence. Or, for those that choose to put their infrastructure in hosted datacentres or public cloud platforms, you may well see the additional processor licence cost passed on to you as a service, as the datacentre providers start to feel the pinch.
Our graphic shows how the new metric will work in practice. I guess the key question, however, is “how will it affect me if I’m already running processors with greater than 32 cores?”. VMware apparently has the answer to that. The press release tells us that customers who buy licences before 30th April for CPUs with 32 cores or more will be able to apply for additional free licences before 29th January next year, to cover the additional requirement for those CPUs. This can be done via your reseller or by raising a ticket in My VMware, though there are, as you might expect, some conditions:
- The servers and licences must be purchased before 30th April – so you can’t just buy licences and stockpile them against future demand. You must have either deployed or be planning to deploy the servers and software and you’ll have to prove you bought the servers prior to the cut off date.
- Customers will have to have active SNS (Support and Subscription) on the licences at the time of the application. Interestingly, VMware does not specify that you have to have active SNS at the cut off date but when you make the request for additional licences. So, for customers running the licences without SNS, you will probably have two options – either just buy the additional licences, or get SNS in place on the ones you are running before you make the request. Whether the latter approach will fly with VMware is not clear at this stage.
- Customers will be charged standard support for their additional free licences when the support contract for the original licences renews.