Yesterday in Part 1 of this article, I wrote about the challenges that we face with DaaS in 2016. Frankly, most of them are the same as they’ve always been, ranging from technical challenges like integration and resource locations to mental challenges like trust in the provider and disbelief that someone else can run the infrastructure better than you can.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
When I finished, I realized how bleak the article made DaaS appear, so if you’ve come back for Part 2, I promise to be more upbeat. Today, we’re going to talk about the benefits and potential use cases that have appeared or matured since we last took a long look at the state of the industry. We’ll start with the benefits.
Benefits of DaaS
The challenge here is sorting through the BS benefits and finding the legitimate ones. If you’ve read anything we’ve written about DaaS in the past, none of these should come as a surprise. Among the BS benefits are things like “DaaS can save you money,” or “You get a fully managed desktop,” or “It’s easy…” Recall that DaaS is just VDI that you pay someone else to build out and manage, but that it’s really only the infrastructure that you’re paying for. You can begin to see why those claims might be less than forthright.
That said, there are real benefits to Desktops as a Service, and depending on your use case, they could make or break the concept in your mind. Let’s take a look:
No huge up-front costs
If you want to deploy VDI today, you have to acquire lots of hardware and software, resulting in a huge up-front spend that you expect to recuperate over time. Whether you want to support 50 users or 5000 users, you’re buying a lot of hardware and software. With DaaS, your up-front costs amount to whatever the cost per desktop is, so where you’d spend tens of thousands of dollars to get a 50 user VDI farm up and running, you might only spend $1,500 to get up and running with DaaS (assuming $30/mo per user).
Predictable costs over time
Not only is the cost to get started lower, it’s also predictable each month. Mutliply the number of users by the monthly (or hourly, or whatever) cost per user and you know exactly what you’re going to spend. There are no surprise costs as you might run into when running your own VDI. For example, no extra storage when you realize you under-bought, or no buying an entire new host to accommodate a small numbers of users just because it breaks your redundancy model.
Incremental scalability and costs
Because of the way DaaS operates, you can scale up and down the number of users you have as needed, which means if you want to go from 300 users to 301, or 325, or 500 users, you can do that knowing exactly what it will cost. The reverse is also true, meaning that if you have a 500 user environment and you want to scale back to 300 users, you can do that and realize the cost savings immediately. With VDI, if you bought enough hardware software to support 500 users, you can scale back to 300, but you’re still sitting on 500 users worth of hardware and software that you paid for.
So those are the primary benefits of DaaS, which line up fairly well with all other cloud services. Now let’s look at how people are using it. I’ve spoken with many providers over the years, and one question I ask all of them is about the primary use cases they see across their customers. Without hesitation, every one of them counts “tired of the cycle of managing VDI on their own” as a reason. Companies want to have remote desktops, but managing them is either outside their wheelhouse or they’re stretched too thin to keep up with all the changes.
Many providers are marketing their platforms as a solution to temporary employees, contractors, and seasonal workers based on feedback they’ve gotten from their customers. Rather than buy and maintain hardware and software that they only use for part of the year, companies are choosing DaaS. These companies may have to work through the technical challenges we talked about in Part 1, but odds are the end users probably only need access to a limited set of applications and data, so it is easier to manage them than it would be for a rank-and-file employee.
Another common use case comes from companies that offer a subscription service to other independent entities, like real estate or insurance agencies. That small group of realtors in might want to use a specific application that an ISV makes, but lacks the expertise to deploy it and make it useable while on the road, so the ISV turns to DaaS providers and sells that back as a service to their customers.
The evolution of IT
You might count this as a use case, and if you do, consider yourself at the forefront of IT. We’ve all heard of the oft-rumored death of Windows. I’ve written about it a bunch of times, debunking the myth that you’re currently behind the times if you run Windows apps (a myth that has quieted down now that VMware cares about Windows). The fact is, though, that we are moving away from Windows apps at an increasing pace.
So why is Windows going to be around? Well, mostly because we have so many damned Windows apps. If your company has 400 Windows apps and you migrated one per month to some non-Windows platform, it would take 33 years to complete that migration! We keep Windows around because of the apps, and until the last one is gone, we’re going to need Windows in one form or another.
That’s where the evolution of IT comes in. Sometime in the future, we’re going to find that we’ve moved enough Windows apps to other platforms that it’s just not worth running Windows on all the desktops anymore. We’ll still have Windows apps to deploy, but maintaining an infrastructure on premises for everyone won’t make much sense. When that day arrives, DaaS providers can alleviate that need for Windows. By moving those legacy apps to desktops hosted in the cloud, you can take the responsibility of managing that infrastructure off your plate while still delivering the apps the end users need.
(By the way, when Windows is used in this manner–that is, just because we have to have it to run an application–we call it “middleware.” It’s not pretty, but it’s why Windows will be around for the rest of your career.)
Here are words I never thought I would utter: Microsoft is being cool about licensing.
2008 Gabe just died a little, and so did many wonderfully funny jokes about Microsoft licensing. Seriously, they made me look like a freaking comedian.
The reality, though, is that Microsoft waited until they were good and ready, then did what we all had kind of lost hope that they would ever do: they announced that they would allow Windows 10 Enterprise desktops with per-user Software Assurance to run in multi-tenant environments. This move takes DaaS from licensing nightmare (or loophole, depending on how you look at it) to mainstream adoption.
As best I can tell this still hasn’t been formalized, even though it was initially announced back at Synergy, but it’s coming. Citrix has partnered with Microsoft to release a XenDesktop-based DaaS platform on Azure, and I highly doubt the service will hinge upon single-user Windows Server 2016 RDSH desktops.
It’s worth noting that this new arrangement from Microsoft is not the same thing as SPLA. You, the customer, still have to bring your own licenses to the provider, but the relaxation on multi-tenancy means that providers no longer have to dedicate hardware to individual customers that want to run Windows desktop-edition OSes. In theory, this could make DaaS cheaper because there will be less unused hardware.
If I had to pick between multi-tenancy and SPLA, I’d pick multi-tenancy every time. SPLA is great, but it really only matters to the smaller companies that don’t have an enterprise agreement with Microsoft. Large companies that EAs have already negotiated a price for Windows licenses that’s probably lower than what the average person pays anyway, so they may not even care about SPLA. Still, if we can get it, that lowers the barrier to adoption for smaller companies, so I’m still rooting for it.
New platform capabilities
One of the main challenges people have from both a technical and metal perspective is the idea that the desktops are no longer on-premises. Not only are companies relinquishing control of the desktops, they’re also distancing themselves from them physically. To address this, we’ve seen a new capability emerge in the last few years: On-premises DaaS. Don’t let the name fool you–it’s really just DaaS–we just have to differentiate the concept somehow for the sake of this article.
On-premises DaaS involves placing the cloud nodes in your datacenter. Some providers do this, like VMware, and Microsoft has all but said they will be able to do it with Azure Stack. This approach of moving your desktops to the cloud while still in your datacenter sounds a bit counter-intuitive, but it can solve a lot of problems.
First, it keeps the desktops inside company walls. You might need this to assuage some doubts or fears about your desktops leaving your property, or perhaps you might be required to keep close ties with the desktops for regulatory or security reasons. Either way, you can keep them close by while also alleviating any data sovereignty issues that DaaS may have otherwise introduced.
Second, you still get to avoid managing the infrastructure. On-premises deployments are usually built on hyperconverged appliances that interact directly with the cloud platform, meaning you manage it as if it was any other cloud offering. The actual hardware itself is assimilated into the cloud platform, and spare maybe adding more hardware at some point to increase capacity, you pretty much just leave it alone.
Third, you get to keep the applications running on DaaS near the data that you keep on-premises. This is a huge deal! Think of all the extra work that it would take to relocate the application data, increase bandwidth, or migrate the application and compare it, well, pretty much business as usual.
The downsides of on-premises DaaS boil down to responsibility. Since the equipment is in your datacenter, you’re responsible for cooling, power, networking, etc., all of which come at a cost. Plus, because the provider isn’t 100% responsible for the entire stack, that leaves open the possibility for finger pointing if something breaks.
If you’re waiting for the flood gates to open with regards to DaaS adoption, that’s not likely to happen. Rather, it’s more likely that it will gradually pick up use cases as they become available. Changes like the ones made to Windows licensing, the on-premises DaaS concept, and the fact that Citrix, VMware, and AWS are committed to it (Citrix is still building their platform) are helping open DaaS up to more potential customers and use cases, so we might see it pick up some steam in the coming years.
In the meantime, there is a lot of interest in the concept of DaaS from new service providers. At least once a month I learn of a new provider that is selling “Streaming IT” (Nerdio), “Untethered Agility” (Paperspace), or some other magical sounding term. Over the years we’ve seen new providers that focus on general use DaaS (DinCloud, Navisite, Quest, and many more), or on specific aspects like security (The Sixth Flag) or GPU (Cloudalize, Frame).
My point is, there is interest out there, but there may not yet be enough customers to go around. It will be interesting to see which use cases and strategies pan out and which ones don’t. The state of DaaS as 2016 draws to a close is neither overwhelmingly great nor horrifyingly bad. You could call it a solution looking for a problem, but that seems unfair. There are many benefits to it that can and will be realized in the years to come, but it’s going to take more companies moving more of their datacenter-based services to the cloud before we can pull off a mass desktop migration. I’m pretty sure we won’t be seeing that in 2017, either.