The Multi-Cloud, Multi-Region Model For Cloud Desktops

By Jimmy Chang, Workspot

Using the public cloud means that uptime is a shared responsibility between your organization and your IaaS provider. To help you remain resilient, your provider usually offers tools like VM shadows and failover zones. However, that doesn’t mean you can wash your hands of the issue.  Maintaining high resilience is a business necessity for end-user computing, so you need to make sure you’ve explored it in depth.

Taking The Blame For Downtime

Traditionally, organizations have dealt with end-user computing as an on-premises workload, addressed either with physical PCs and workstations or via do-it-yourself VDI managed by an in-house IT team or a managed service provider (MSP). With the goal of zero downtime, IT organizations try to plan for all possible outages. This is just good business continuity planning: anticipating various disruption scenarios and then determining how each will be addressed. Most organizations have at least a basic plan in place for their on-premises infrastructure, including backup and disaster recovery (DR) procedures, with processes and playbooks honed and documented by IT. 

COVID-19 turned this paradigm on its head. Almost overnight, mission-critical computing expanded to include end-user computing for a huge percentage of remote workers. Suddenly, the concepts of downtime and productivity loss became acute and even existential for many businesses, yet IT teams found that the existing playbooks for avoiding business disruption no longer applied.

A VPN approach was one possibility to address the sudden shift to remote work, but many organizations experienced performance bottlenecks and security concerns – especially as it became clear that supporting remote work was likely not a short-term challenge. Other companies expanded their use of on-premises VDI, with varying degrees of success. And many companies looked to the cloud as a long-term solution to support a flexible “work from anywhere” strategy going forward. However, if you move all your critical services, including end-user computing, into a single cloud region, then you have simply moved the same outage risk from your on-premises infrastructure to a single cloud infrastructure.

The Multi-Region Approach

Since resilience is the goal, it compels organizations to think about a multi-region cloud strategy. You have a certain amount of control over many problems that could arise by centralizing desktops in the cloud, but a particularly challenging risk is the possibility that an entire public cloud region could go down. Though this is unusual, it can happen – and it has happened. Organizations moving desktop workloads to the public cloud need to plan for this eventuality just as they have planned for other outage scenarios. 

If you have a cloud desktop solution that resides in one cloud region, and that region goes down, your users can’t access their cloud desktops, and that means the business is on life support until service is restored. For some industries, such as financial services, the cost of this lost productivity can be millions of dollars per hour. This is an essential consideration as you think about the return-on-investment of the solution options you’re evaluating.   

Organizations expect to measure Recovery Point Objective (RPO) and Recovery Time Objective (RTO) in days. If you’ve chosen a solution that can also quickly fall over to an alternate cloud region if the primary cloud region goes down, you’ve achieved the gold standard of resilient end-user computing. Now, you can have an RTO measured in minutes and an RPO of <24 hours. In addition, some solutions offer “stand-by” cloud desktops that can be activated by your IT team in minutes when disaster strikes.

Resilient End User Computing

Companies with high availability requirements and on-premises data centers would typically use redundant networking and data centers to reduce risk. Traditionally, end-user computing wasn’t part of that multi-data center strategy because VDI doesn’t scale well horizontally. It’s too expensive, complex, and difficult to maintain across multiple data centers.

However, as cloud strategies mature, many organizations now require a multi-cloud approach because they must consider the possibility of a provider-wide outage, and they need to avoid being overly reliant on a single provider. Fortunately, in the era of the public cloud, deploying cloud desktops across multiple cloud providers – if it can be managed from a single console – can dramatically simplify IT’s ability to align the goal of flexible end-user computing with broader organizational goals for resilience.

Starting The Multi-Cloud Switch

It’s quickly becoming a standard practice to select cloud vendors based on your priorities and assign some workloads to one cloud vendor while deploying other workloads in another vendor’s cloud. Solutions such as Google Anthos and Azure Arc are great examples of core technologies that make it easier for organizations to execute their multi-cloud strategies. Multi-cloud end-user computing must be simple for IT, too.

Deploying cloud desktops across multiple cloud regions or cloud vendors improves resilience and gives them better RTO and RPO than on-premises solutions. But the last thing IT needs is more complexity, and most virtual desktop solutions are highly complex. Instead, look for a desktop-as-a-service (DaaS) platform that enables backup to an alternate cloud region, which protects against regional cloud outages.

If you want to use multiple clouds for even more resilience and to double-down on low-latency desktop access, consider solutions that centralize management into a single console. Otherwise, it’s complicated to keep track of global user experience and stay ahead of any issues that may arise. Even the cloud incarnations of legacy VDI, where the broker is hosted in the cloud, can’t deliver this kind of simplified management. It’s an underlying architectural deficiency that prevents them from achieving this for customers. 

Design For Resilience

The pandemic forced a quick shift to remote work, and many organizations soon discovered that the method they’d chosen had latency, scalability, and security issues. Cloud desktops are a great option – until the cloud region they’re in goes down. With resilience at stake, a multi-region or multi-cloud DaaS approach to cloud desktops is a safer approach. It also reduces complexity and workload for the IT team. Consider the best practices above to develop a cloud desktop approach that best meets the needs of your end users.