Over the years VDI has been a great tool for promoting workforce mobility, centralized management, BYOD, data governance, data security and many more use-cases in enterprises. However, it is important to recognize when times are changing and try to adapt yourself to the new reality. VDI is morphing, requirements are shifting and I think it’s time to pay attention to both the changes in the market and technology.
How did we get here?
Most of my readers would know that I am not a prominent supporter of DaaS (Desktop-as-a-Service). However, the reality is that any enterprise can have a successful DaaS implementation simply creating a self-service portal where users can provision virtual desktops, as needed, automatically, without requiring human interaction.
DaaS self-service portals should be an easy-to-use user interface that abstracts infrastructure from users and operators. Underlying technicalities such as datastores, resource pools, desktop images and other complex structures should not be something that VDI operators or users should deal with in order to manage virtual desktop deployments.
Assuming this convention is correct, DaaS is nothing more than VDI with a self-service portal. I might be oversimplifying a little as I am not considering aspects mostly applicable to service providers, such as multi-tenancy and charge back.
Despite having fondness for the simplicity that DaaS solutions promote, I simply cannot see how organizations supporting hundreds or thousands of applications can move those to a presentation layer solution and present them into virtual desktops hosted in an external infrastructure provider or cloud. Furthermore, some applications may never be suitable for this type of technology.
While VDI has seen success in terms of adoption and growth, DaaS has not delivered on the hype. DaaS was all about only hosting the desktops in the cloud while keeping apps, file servers, AD, etc. on-premises. That did not work that well because apps and data need to be close to the desktops. DaaS deployments struggled and failed because of that.
But with new architectures and public clouds, it is possible that some of the workloads/apps are also moved closer to desktops. Now companies that want to use public cloud view them as new “branch offices”. So, move important apps/files also to the cloud, move AD to the cloud, and buy a good dedicated MPLS line to the cloud. DaaS requires a much more careful planning but that doesn’t mean that it cannot work for smaller organizations.
The birth of SaaS-VDI… or SVDI
I coined the term, but you may be able to come up with something better.
The beauty of DaaS is the simplicity and self-service driving the user experience, the OPEX consumption model, and the abstraction of infrastructure complexities. It was only a matter of time before people understood that the best way forward is making DaaS very simple and functional, yet providing true access to all enterprise applications and data sources.
Prominent VDI solutions are complex when it comes to infrastructure components – managing these systems components appropriately and with all the required disaster recovery procedures and workflows can be a daunting task even to the most experienced operators. I personally don’t take anything away from those VDI solutions as they are very mature and efficiently deliver on what they are selling.
The tendency is to eliminate complexity and the need to manage and maintain on-premises VDI infrastructure components, leaving only virtual desktops and applications located in the datacenter whereas everything else is provided and managed by the SVDI vendor via self-service portals. In the age of SaaS why would anyone want to concern about VDI infrastructure and workflows, such as provisioning services, connection servers, backend SQL databases, license servers, security servers, server patching, gateways, multi-site management, PODs, Cloud PODs, database backups, database replication for failover, DR for infrastructure components, types of desktops etc?
In all honesty, if enterprise apps, data and ancillary services are readily available in the cloud where virtual desktops are hosted, I blindly recommend a DaaS solution. Regrettably, that’s not the case for the large majority of organizations – some have over 1000 apps to be managed.
Going back to the matter – Although some infrastructure components are still required to connect the on-premises virtualization infrastructure (vSphere, XenServer, AHV, Hyper-V) to the SVDI cloud service, the number of component considerations and point of failures are greatly simplified. The cloud management solution is now responsible for all VDI components, multi-site management, backup and disaster recovery workflows across all sites and regions. This is somewhat analogous as to what hyper-convergence is doing, removing complexity of the datacenter – but in the HCI case the infrastructure components run on-premises.
In this new model enterprise IT is responsible for managing and maintaining the virtualization stack while VDI components are a responsibility of the SVDI vendor. The approach implements the best of both worlds, the cloud-based management with touch-less maintenance and an on-premises delivery of virtual desktops and apps.
Taking a step further, combining SVDI with hyper-convergence you now have a self-sustaining and scale-out solution that is predominantly touchless from an infrastructure perspective. The new approach not only has technical and operational benefits, but also a subscription consumption model for the VDI infrastructure.
VDI vendors with different products for on-premises and cloud will fail if their products are not properly integrated and able to operate on-premises and on the cloud – VDI and DaaS must be seamless to administrators and users.
Citrix is heading in similar direction with their Citrix Cloud (formerly Citrix Cloud Workspace). Citrix enables management of on-premises Citrix delivery components from their web-based portal and they can also deliver services on public clouds. I can only presume that Citrix is also looking to deliver a turn-key solution integrating their cloud-portal and hyper-converged stacks with minimal requirements for on-premises infrastructure components, effectively abstracting all the complexity.
Another new vendor on the cusp of this approach is Workspot. Workspot partner with hyper-convergence vendors (Nutanix, Atlantis, ScaleComputing, Simplivity and others) to deliver a turn-key solution that spans on-premises and cloud. In their case, the VDI operator only needs to drop a virtual appliance on-premises – pretty much everything else is done via their cloud portal.
There are unquestionably components that would still need to be managed by IT teams, including operating system, profiles, VPN and user data that could have further impact on the overall solution design. It is important to analyze the pros and cons for each approach, including all technical and financial aspects.
It is also important not to lose focus of why VDI exists, as one of my twitter followers wisely said recently “VDI is not important, VDI is a simple way to work on what really matters to users – applications and data”. The key is that in the next 5 years, VDI will also transition from a workloads like “Exchange” to something like “Office365”. The TAM of VDI will increase if vendors can further simplify and make VDI more like “Office365”.
I am in favor of IT simplification and complexity reduction, and I look forward to see these solutions maturing overtime. I also know that this type of technical and consumption model disruption may have a difficult time getting acceptance from hard-core VDI operators and architects. However, if you have a deep look on the market and technology and consider the current direction I’m sure you’ll agree that it makes sense to eliminate datacenter complexity.
Disclosure – The company I work for is a partner to Citrix and Workspot. I joined Workspot’s advisory board in late 2016.
This article was originally posted on LinkedIn. For comments, please go to https://www.linkedin.com/pulse/end-vdi-we-know-andré-leibovici