VDI is dead; long live VDI! Here's why

Martin Banks Profile picture for user mbanks February 2, 2024
Former Citrix CEO Mark Templeton argues that available technologies and infrastructures are coming together to create a new, and potentially much bigger market opportunity for a new-look VDI.

Image of colourful play stones connected by lines on a white board

Back in the day, systems utilizing Virtual Desktop Infrastructure (VDI) were widespread and most often were sourced from that market’s big gorilla, Citrix. Business was good, mainly because the very nature of VDI suited it best to the needs of enterprises and organizations where security and management control were most important, such as financial institutions, healthcare bodies and large government departments.

Citrix did well enough out of it as a business, but its last available financial results in fiscal 2021, showed a one percent revenue decrease over the previous year, at a time when the pandemic might have been expected to provide some boost due to the rise of working from home. This indicated that the implementation of VDI it was offering was no longer hitting the sweet spot of user needs, at a time cloud-delivered services were starting to offer more flexible and cost-efficient alternatives to users. 

That was when Vista Equity Partners and Evergreen Coast Capital Corporation, acting through affiliates, moved to acquire Citrix for a total of $16.5 billion, and put it together with Vista’s existing major IT company holding, TIBCO, to form a new business, Cloud Software Group (CSG), a move that indirectly opened the way for Workspot to move more front and center in the VDI marketplace.

This, according to Templeton, is because both companies in CSG have reached the point of such maturity – both in terms of marketplace and available technology – that only existing customers having long-term needs will continue using the technologies and products of the CSG constituents, while the available marketplace shows little or no scope for future growth.  The alternative, therefore, is to move them into what Templeton calls `the milk shed’:

This is the reality of the playbook that they're running. To be rude, if they're running the milk shed playbook, they are cash-counting a bunch of software assets. It shouldn't be a surprise to anyone. It means they focus on premium customers, and if they use partners they focus on premium partners. And then they let the Darwinian process work with everyone that doesn't clear the premium partner bar or customer bar. Then, of course, you streamline, which is a euphemism for elimination of a lot of things that sustaining growth businesses need.

The result is that, for those premium companies that feel they have little room for growth in their own business, the availability of established legacy hardware, applications and services is seen as the best, most economic and most secure option for their futures. The corollary of this, usually, is that the life of the customer businesses is relatively short as their own markets move to new suppliers and new business models. The financial institutions are an obvious example of this in action, with even retail banking now being almost unrecognizable when compared to the old ways of cheque books and high-street bank branches.

Templeton sees this as one of the inevitabilities of doing business and a legitimate business strategy as it creates legitimate value for the shareholders. It can also create a breathing space for customers, long committed to applications and services that have worked well and still fit the current requirements. They get time to think about what should come next. But that thinking time does come at the cost of a hefty 'pound of flesh'. For the duration, there is no alternative.  

It can only get worse, as well, as 'the farmers' are also cutting costs, such as staff involved in new product or new market development. Templeton notes that Citrix is shedding a further 12% of staff, following a 15% cut last year:

I'd say what they're doing is rational, when you're running the playbook. And it works. It's very successful. But when you're on the side of the market, where you're trying to drive innovation, growth, and move from legacy platforms to modern platforms, that's a whole different story. And that's where Workspot is as a company.

The comparison he uses involves some well-known names from the industry’s past. Siebel, BMC and PeopleSoft; all huge juggernauts in their time and all implemented in the customer’s data center, where they were costly, and complex to implement and run. Each was supplanted in the users’ hearts and minds by, respectively, Salesforce, ServiceNow and Workday. 

All three are cloud-based and cloud-delivered services rather than data center applications, and this Templeton sees as an important part of the milk shed playbook. He sees a significant divide between those staff that understand the data centre/applications operational model, and those that understand cloud services such as SaaS. He includes himself as an example:

Data center engineering teams understand how to build for data centers. Cloud is a sideline for them and what it takes is way more than they've done. So their cloud initiatives are not actually very successful, and this is across the board for the legacy guys. This is actually the reason Workspot exists because back in 2010 or 11, Amitabh [Sinha, CEO and co-founder of Workspot] was running our big Zen app group (at Citrix) on the engineering side, and he came to me and said, 'We need to build, go to the cloud’.

That was our strategy, but we needed to start over, especially with the talent, because the cloud services programmers are not the enterprise data center programmers we needed, and we didn't have, that cloud competency. And, you know, I didn't understand, I certainly didn't agree. So he left the company, and he said, 'I'm going to build a cloud native solution, long before it was even possible to do it. We never did that at Citrix. They never did that at VMware. Their strategy has been to lift and shift their enterprise data centre technology into the cloud.

Templeton's reaction to my suggestion that Citric and VMware are losing their grip on the VDI market just at the time it is reaching a point where it can actually deliver some real end user advantages, particularly in the area of data security, better sustainability and highly flexible simplicity is simple:

Correct. When you drop that premise in the context of a world that's more volatile and uncertain, and more ambiguous than ever, you need a generational change.  Tectonic plates are shaking the core of classical distributed computing. Think of it as a bigger thing: a Desktop-as-a-Service, streaming enterprise desktops, apps, workflows and data. You know where that full stack is managed, where they can control the costs, the security, where and how things are stored.

Where does VDI’s new life come from?

VDI has been dying for a good few years, collapsing under the weight of its own complexity. What has changed is that much of that complexity is, in practice, meat and drink to the cloud service providers, and in particular the hyperscalers. Ten years ago they were a suspicious rarity, but now they are everywhere – or at least their services are.  That is where the complexity is hidden from the user and that is the key nature of the service provided. In addition to complexity, performance, reliability, sustainability and operational flexibility are also provided. If it can be defined in code, it can be delivered by cloud faster, more reliably and instantly modifiable as and when required, be it to one user or many thousands. 

Templeton turned to Sinha, to talk through what the company is now offering, and why they feel VDI is being reborn:

VDI is dead because it was built for a single data center in a single region for a single customer, and every customer had to do all the hard work to get their own VDI going. But end users could be distributed all over the world and were using slow networks offering poor latency, so the user experience was bad. Everybody suffered it and users hated it. But it offered security. If you`re in financial services, healthcare, or had to protect data, there was just no alternative to Citrix for those customers. So people used it in spite of these problems.

Even that claim has recently been downgraded. Last November the tech press carried stories about Citrix Bleed, a flaw in the Citrix code that allowed hackers to attack multiple companies. A patch has been produced by Citrix, but it has to be noted that the milk shed business model includes the reduction of expenses, such as the staff needed to develop such patches. So its ability to defend customers against future flaws and attacks has now to be open to question.

Now, with Azure, AWS and GCP available, Templeton claims there are about 250 data centers available to every customer, regardless of location. So a business with a New York HQ but many users in the Philippines, you can deploy in Philippines and Singapore and Taiwan. It is a case of pick the data center closest to that end user, for most of them will now have reasonable bandwidth networks at home or at the office. The Workspot environment allows users to just consume the service and removes the need to spend 9 to 12 months designing and deploying dedicated infrastructure in a data center.

He adds: 

And it scales easily. I can start with five users, or 50, and I can go up to 5000. And it's really very quick to get up and going because it is designed to simple and scalable, so users can sign up each use case in their business as and when it is ready. There is no all-or-nothing option.

Divide and conquer

The second part of the equation is a fundamental change in the way the operational and management code is written, which meant a complete re-think of VDI from the ground up. In a VDI system there are two key elements, the control plane and the data plane, and in traditional Citrix environments they are inextricably entwined – so everything has to be in the one system. With Workspot, Sinha’s key development has been to separate these two.

As the names suggest, the control plane is about managing the complexity of, say, the connections of user desktops to the central system and the management of their activities while connected. This, however, is more about set up and oversight rather than hands on activity on each desktop, and is best centrally located at the main data center. The requirements for the data plane are, however, almost the exact opposite. Each user can, if necessary, locate their own data plane locally, say at one of the 250 or so data centers readily available to them, but under the full control of the control plane. If the user has its own suitable resources, including an existing data center, the data plane can be located there.

Templeton explains: 

All cloud services have had to do the same innovation, to separate the control plane from the data plane. And once you can do that, it's a core networking concept. Once you do that then we can take advantage of all of this stuff, because then you really don't care where your data plane is in the world. And we can have a centralized control plane which can manage millions of data planes around the world.

As it is SaaS-delivered, Workspot is easy to get started with and can be re-configured, expanded, or reduced, at any time. They can also carry accessing data sources as they always have. No special provisions need to be made, or data relocated, says Templeton: 

We have had customers that have gone live in an hour. Whether customers will take longer than that depends on the sensitivity of the customer and how they bring new services into their organization. The most difficult conversation for us is with the networking team, which is when they ask, 'Can I trust your service in my data center? I want to validate its security profile before I let it in’. And that's really the longest part of the conversation we have.

My take

This re-design should provide a platform capable of handling the vast majority of business workloads, with only those requiring the most outrageous commitments of compute resources liable to being beyond the pail. In addition, because the classic VDI model – all the compute and data is within the confines of the data center and all that its transmitted to and from the user client is pixels – is maintained. It means that any individual user can use virtually any device as a client, from the latest laptops to old PCs and, course, smart phones and tablets, from anywhere that an internet connection is available.

In addition, security is far greater than with a standard PC-Internet-data centre connection because even if a logged-in client can be used as an access point for an attack, the attack has to get past the superior defences of a hyperscale data center. Also, there is an increasing important contribution to sustainability and corporate carbon footprints, as the gross amount of data generated and stored when using VDI is significantly less as file duplication is greatly reduced, which in turn reduces the energy consumed by the process.     


A grey colored placeholder image