Yesterday as you guys noticed, I posted about using Azure’s Windows Virtual Desktop (WVD) to broker workloads that are not running on Azure (i.e. on-premises, other clouds, etc). This created a massive discussion on Twitter about its legality, supportability and so on.
But what surprised me the most was some people’s reaction. Many people do not seem to understand why it was important to point that out in the blog post and the valid scenarios where WVD can be leveraged.
The reason this reaction from several people took me by surprise is quite simple. Some suggested that Microsoft would be certainly blocking this. So instead of coming up with a way to monetize on this particular scenario (allowing people to broker workloads anywhere using WVD), Microsoft would consider implementing ways to prevent this, to make sure customers don’t do it.
It is worth to point out that as of May 2020 there is nothing on the WVD EULA/Service Agreement that clearly states that you cannot run Windows Server or Windows 10 Single User workloads outside of Azure (i.e. on-premises). For Windows 10 multi-user they do mention that but not for other workload types. From a technical standpoint, any workload can be brokered by WVD regardless of where it is running. But now, some people do seem to think the best approach is for Microsoft to devote time and effort from its own RDS engineering team to block this. This would not be a good approach for several reasons. One is wasting resources on this instead of improving the service (the ‘Spring Update’ is a great example of what can be done when you want to improve an offering). Another is, if customers are doing it, it is clear there is a need/demand for that.
Wouldn’t it be better to simply work around the RDS licensing (or other licenses) to come up with something like ‘RDS CAL Plus’ or ‘Microsoft 365 E5+’ where you charge a bit extra but then allow the customer to run such workloads anywhere they want? Or simply determine if the workloads are on Azure and if they are, no charges for WVD usage but if they are outside of Azure, implement a per hour/per user model and charge customers like any other Azure service.
From a technical standpoint, some of the arguments to tie you to Azure are not valid IMHO. Monitoring for example, anyone can use much better monitoring tools to monitor everything (i.e. ControlUp, UberAgent). The same can be said for several other potential ‘drawbacks’ you may encounter by running such workloads somewhere else.
But given the reaction I have seen, some people do believe that finding alternate ways to better license this is not the way to go and actually preventing customers from doing it and locking them further down into Azure is more appropriate (and potentially, according to Microsoft, better to you the customer – after all Microsoft knows better what you need and want, right?).
Given the massive debate this caused on Twitter, it seems customers do think WVD is a valid, solid offering and one they would love to be able to use to leverage their on-premises or non-Azure hosted workloads.
I am not sure what other folks in the industry think but this, on top of the whole Office 365 licensing limitations that prevent you from running it on different clouds in certain scenarios (i.e. when you subscribe to these already but want to leverage AWS workloads), clearly shows the direction Microsoft is taking.
I can certainly understand a company willing to increase its market share in a certain area (Cloud) and to make its offering (Azure) as attractive as possible. But making it attractive with restrictive licensing and with severe lock-in, making it illegal to leverage certain services on different providers is not a great way to do it.
I hope this post opens your eyes to Azure and WVD and makes you rethink your strategy. I see suppliers like Microsoft as partners. This is far from a partnership IMHO. It is a one-way road where all it matters is Microsoft, where you, the customer, is irrelevant.
You have been warned.
And yes, if Microsoft does change the WVD licensing and starts to block you from using it outside of Azure you can blame me.
If I vanish all the sudden, you know what happened with me.
CR
I agree with them.
Fair enough mate. Would be great to understand your reasoning behind (also on the subject blocking vs finding ways to charge for that for non-Azure workloads)
They do charge – If I’m understanding the licence model correctly, to be compliant when running a workload on-premise, you would require an M365 Biz Premium licence for the user, PLUS you would need the Server OS licence and CALs, or desktop licences plus VDA/SA.
I cant understand why people this is a problem, seems like a perfectly valid use case and an opportunity to grow the market. I have a number of clients running legacy Horizon or Citrix, who would love to swap out the broker layer for cloud based PaaS, and since Microsoft have a queue of people waiting for NV4as desktops that they can’t service, I don’t see the problem.
I cannot find it anywhere that says certain workloads can be brokered by WVD AND be on-premises. No matter the licensing. If you have a link to some sort of FAQ from Microsoft that explains that, it would be awesome. And yes, I have no clue why the pushback from many. It is all about flexibility and being able to go to the cloud at your own pace. And to go to any cloud for redundancy/DR purposes.
CR
Sorry – my meaning was that all that licencing would be required just to be compliant with MS licencing policy for both on-prem RDS and WVD.
i.e – if you want to do it that way, you are already paying twice (M365 for WVD licencing, CALs / SA for On-prem licencing). So its not as if MS are losing licencing revenue from this edge case exploit.
Azure resource costs are a different matter however, and I’m sure these are factored into the low entry cost for WVD service.
Oh look…
https://azure.microsoft.com/en-us/updates/azure-virtual-desktop-for-azure-stack-hci-public-preview-is-now-available/