Virtual desktops are not new, of course. In fact they are now classed as one of the classic legacy applications that millions of people know and have worked with, and which is now probably on the gently slide down to its pensioning off.
Well, maybe not. In the opinion of Nikola Bozinovic, General Manager of Desktop Services at Nutanix, the time for VDI is only really just starting, and what has gone before is not just a prelude, but a tune set in a time warps of endless repeats.
But all that, he predicts, is about to end. The availability of higher performance, hyper-converged compute resources, coupled with new software technologies based around the idea of the browser as the operating system, plus the early availability of 5G communications systems, is opening up entirely new opportunities for making seriously good use of the core VDI approach – maintaining high levels of security by only ever streaming pixelated representations of tasks being run on applications running in the cloud.
Bozinovic’s view is the VDI is dying only if one feels that the underlying technologies- the software itself and the hardware and systems platforms on which it has run - are incapable of change or development. If that is the case then yes, it is well past its prime and well due for that pension. But he takes the view that both sides of the equation are in the process of major upheaval, and that the outcome for VDI is exceptionally good.
The emergence of hyper-converged architectures from the likes of Nutanix has already changed the underpinning `mechanics’ of running VDI services, allowing them to be scaled out and run on far more powerful, faster platforms than ever before, he argues:
VDI is probably one of the most complex things. You need a lot of server-side power. We've been used in the last few 15 years to it not having graphics yet, it does not have GPU. Your phone has GPU. your PC has a GPU yet 95% of world's VDI is without the GPU. Because it was complicated to virtualize a GPU, it was costly data.
He sees bandwidth as another hindrance, which with 5G around the corner is something that will have change rammed down its throat, together with security, especially as we move into a multi device world:
Every year for last 15 years has been called 'the year of VDI' - and they're still asking that question. So obviously we didn't have a year of VDI, and when the iPhone launched we were not asking because mobile phones were here.
The move to cloud services has only accelerated this process, and this has made it clear that the software itself has not kept up. In his view it is not multi-tenanted, it is not very scalable and not very secure.
Trying to address these weak spots and the new potential for VDI are the start points that have led to the arrival of Xi Frame as part of the Nutanix platform.
One of the key start points was the launch by Google of its game-streaming service, Stadia, with 4k 60 frames/sec video, running seamlessly in any browser, and that integrates with other cloud services.
As Bosinivic pointed out, if you look at it, it's a VDI, and because it was about playing games it was going to get a lot more users. That became the model for Xi Frame: browser-based and with everything as a cloud control plane. They key difference would be that there would be a much higher commitment to security issues than found in a game playing system. He sees this as a shift towards sort of managed service on the desktop:
VDI has never penetrated to more than 10% of business. So the market is there, a lot of people are on Windows, but now people want to be on Chromebooks. People want to be in different endpoints. And I think that's what VDI has going for it because it is throwing them independence from any architecture. With Xi Frame it is endpoint agnostic.
By approaching the issue with the idea of the browser being the operating system, he sees Nutanix taking the opportunity to rise to another level of abstraction, where applications such as movie editing – or anything else requiring 4K-level graphics – and the ability to provide fast and efficient file management He sees the browser now being the most resource-demanding application a single computer can run, especially when they have too many Chrome tabs or edge tabs open. Yet the capabilities it provides is significant, and growing:
And I think we're just scratching the surface, and that is the freedom you can get by having an endpoint browser on any device. Is this a tremendous opportunity to go there?
His answer is `yes’, if only based on the fact there are thousands of Windows applications that can be migrated and then delivered, with the inherent security afforded by VDI delivering no data but only a pixelated representation of the data and the user’s actions, onto any endpoint device using 5G for the communications. But he sees that as almost a side issue, for he sees the next generation of apps as being far more important, not least because of what he sees happening around them at Google:
They're building a Linux operating system. And they give them unity and all these other tools to build the new games. So the concept is solid from security perspective, from device and dependency. Like, it really works.
He also sees potential for VDI now having the muscle to open up new applications areas, such as becoming a data transport mechanism between applications. Indeed, he would name Zoom as a new application of exactly this type:
If you look at what zoom is doing with video do you call it communication, or its collaboration, because you're screen sharing. Many times I give control to someone else. I can hand it over because I own the meeting. And when I close it, it's gone. Imagine if we were working on a system that would seamlessly change owners and data would follow me.
So will 2019 be the `year of VDI’? Probably not, but that is only because it will take a bit of time for users to get into the wider context of 5G, cloud delivery and multi-cloud, multi-client systems. And here, security is likely to be an important issue. For while the same architecture can be addressed using a variety of different communications approaches between devices, applications and delivery systems, each one represents a direct connection of some kind, and therefore a possible attack vector for the hacking community. By using a pixelated representation sent to and from a (presumably) well-protected server, rather than the data itself, any number of different end-point clients can be used in what would certainly looks like far more safety and security. So give it till 2021 or so before it will be possible to really say 'the year of VDI'.