For many organizations, migrating apps and building new apps in the cloud is inevitable. It is unlikely that any known force could slow down the pace of cloud migration and innovation. Likewise, because of the glut of cloud providers in the market (and their various strengths and applications for specific purposes), multi-cloud is also an inevitability for many organizations. But inevitable does not mean smooth sailing: there are operational, security, and organizational challenges involved. It is important to think through these challenges and how you will surmount them if you want to build an application platform across multiple clouds and deliver a consistent operational model, a consistent developer experience, and a consistent model for security and governance.
The origins of multi-cloud date back further than most people think. By my calculations, its genesis can be traced back to 1998. That was the year that Google was born, the year that Microsoft became the biggest company in the world (with Windows 98 flying off the shelves), and the year that — in sunny Palo Alto, California — VMWare was conceived. Fast forward 23 years, and these are now three of the six global cloud hyperscalers, according to Synergy Research Group (see diagram). Alongside IBM, Oracle, and AliCloud, they dominate the public cloud market and are all delivering VMware-based services on their platforms.
Figure 1: Cloud provider market share trend
VMware can work with all of these clouds, as well as with 4,300 other independent VMware cloud providers operating across 150 different countries, serving more than 150,000 end customers by hosting over 10 million virtual machines.
For most medium and large organizations, adopting a multi-cloud strategy is already a fait de complis — a fact that has been perfectly articulated by my colleague Kit Colbert in his series of articles on multi-cloud strategy. Kit’s focus in these articles is to address the changing tide of enterprise application development and how this is driving a multi-cloud strategy for many organizations.
Figure 2: Almost $5 trillion worth of cloud innovation
Multi-cloud as a philosophy/strategy is more about exploiting the unique capabilities of different clouds, rather than going “all in” with one specific cloud provider. Each public cloud embodies a unique interaction model with different ways for end users to authenticate and to then consume, request, or modify a service or application. Each cloud also comes with a unique set of API surfaces. For most businesses, this becomes a balancing act of weighing the benefits of adopting multi-cloud technologies against the burden of supporting them.
Another key challenge for organizations that employ multiple public clouds is the operational overhead of maintaining common governance, security, and compliance models across this complex multi-cloud environment. This is such an important aspect of running a modern business. Many organizations are realizing that they would like a single, more unified approach to meet these often legally required controls. In addition, maintaining skillsets in the fast-changing world of cloud and the ability to train teams to operate multiple public clouds puts a lot of pressure on cloud operational teams.
Consistency is the cure for complexity
To simplify multi-cloud complexity, organizations are beginning to adopt a hybrid cloud model. I discuss this topic extensively in my free new eBook, “Modern Infrastructure with VMware Cloud on AWS.” The key differentiator here is that whereas the term “multi-cloud” aims to describe how organizations use multiple cloud providers to meet diverse technical and business requirements (illustrated in Figure 3), the term “hybrid cloud” is generally used to describe the combination of private and public cloud infrastructure — where “private” is either on-premises or hosted in a colocation facility and “public” is provided by one or more hyperscaler clouds. In a hybrid cloud scenario, common management and orchestration tools are used to deploy workloads and maintain the balance between the two, as illustrated in Figure 4.
Figure 3: Customer multi-cloud strategy
Hybrid cloud is attractive because many businesses would, ideally, like to modernize at their own pace and to find a way to do so, starting with their own datacenters. They would rather do this than be under pressure to move everything to the public cloud as quickly as possible, which can be very risky, from multiple perspectives. Also, many organizations are looking for a more granular step-by-step journey towards modernizing applications, while gaining many of the benefits of running in the cloud and maintaining workloads under a single governance of control.
A hybrid cloud delivers on this idea by creating a bridge between your own on-premises environment and one or more public or hosted cloud environments. This model often includes employing public cloud resources for regular or occasional bursts of compute and/or storage capacity, for instance, adding capacity on demand to provide additional resources for end-of-quarter batch processing or seasonal bursts in website utilization.
Figure 4: Hybrid cloud architecture
This hybrid cloud experience might also include the ability to extend networking rules from on-premises to a public cloud and by providing a common management-and-orchestration platform, so there is no need to learn a whole new skillset. You can continue to use the same tools with which your operators are already familiar. This can also help to address the gap between the business and IT, as IT can remain in control using all the same procedures and management processes they have worked with previously, and the line of business can get to the cloud faster, without having to re-architect the application.
Apps run businesses
As I’m sure all readers are aware, it is impossible to address any aspect of cloud adoption or use of multiple clouds without considering the impact on existing applications and future application development. Enter stage left, the VMware Tanzu Portfolio.
The goal of VMware Tanzu is to deliver a modern application platform that can help VMware customers transform business outcomes. When it comes to Tanzu, it isn’t just about the IT department. VMware Tanzu portfolio products — including Pivotal, Heptio, Bitnami, and Octarine — provide complete end-to-end solutions for customers to run, manage, and build their cloud-native applications on a single consistent platform that can operate across VMware Cloud Foundation, vSphere, public cloud (VMware, Amazon EC2, and Microsoft Azure) and edge environments.
Figure 5 illustrates one such architectural approach. Tanzu Kubernetes Gride (TKG), which provides VMware’s Kubernetes runtime (an opinionated and supported version of upstream Kubernetes), delivers the consistent application platform alongside existing virtual machines running on VMware, as well as alongside public cloud instances of the same runtime. Tanzu Mission Control (TMC) provides the operational consistency and governance. It centrally operates and manages all Kubernetes clusters and applications at scale, across all cloud endpoints, both VMware and native.
Figure 5: Multi-cloud application modernization with VMware Tanzu
As illustrated in Figure 5, this approach allows the customer to have a single infrastructure platform, where traditional enterprise apps and cloud-native apps can run side by side. This approach allows operational teams to consistently manage and operate Kubernetes clusters and applications across clouds at scale, while also enabling developers to have self-service access to cloud endpoints for running and deploying applications, making them more agile and efficient, with increased productivity. This in turn gives developers the ability to deliver high-quality apps that solve business challenges and generate new revenue streams.
This increased visibility into the software-defined datacenter can allow operational teams to gain access to the required metrics to optimize costs and maximize uptime. Security teams can also benefit from this increased visibility (in addition to the consistent approach), by allowing the management and configuration of applications through a single policy engine, delivering better governance across the entire infrastructure.
Additionally, this ability to provide a “one-platform, any-cloud” approach for both traditional apps and new cloud-native apps is helping customers across the globe focus more on application modernization, whether it is in an existing on-premises private cloud, public cloud, or a combination of both, with a “code-once, deploy-anywhere” mindset.
More control, more simplicity
In summary, this homogeneous approach to cloud and application development facilitates far better control of data and simplifies a wide range of operational factors, such as access, performance, availability, security, data protection, and governance.
Comparing native public cloud services, such as Amazon EKS or Microsoft AKS with TKG running on a VMware software-defined datacenter, is not easy and is a bit like comparing apples with oranges or beer with wine. Quite simply, it comes down to what your priorities are and what your end goals need to be from a business perspective. While it might be possible to do both a commercial and technical comparison between workloads running on Amazon EKS or Microsoft AKS with those running on TKG, that is rarely the full picture. It comes down to tactical and strategic business priorities.
For instance, if flexibility, choice, and control are key requirements when it comes to workload placement, then VMware can offer that, along with a consistent platform across private and public cloud, which makes it easy to move in and out of cloud infrastructure. This is often one of the key concerns for many companies when moving to the cloud. This is due to the proprietary technology deployed by public cloud vendors. The more native innovation and evolution that takes place, the quicker this leads to maximum lock-in and minimum flexibility.
Also, cloud-native services will typically benefit from the cloud economic model, where you pay per second for capacity. Under the VMware model, you pay per host, which, despite the move from CapEx to OpEx, is the same economic model as an on-premises solution. Therefore, the consolidation ratio is key. The cost of cloud-native workloads is consistent — it doesn’t change based on how many workloads you deploy. However, in the VMware model, because the host cost is fixed, the cost per workload decreases linearly the more workloads you are able to consolidate onto a single host. This approach means the insight and visibility highlighted earlier in this article, across all endpoints of a multi-cloud architecture, is key to building a strong business case.
Finally, unlike native public cloud services, which will provide a commodity platform that is likely less compatible with existing applications, VMware Cloud and VMware Tanzu provide a trusted, high-performance, and resilient hybrid cloud platform built on the trusted foundation of VMware vSphere, vSAN, and NSX, supporting both existing workloads, as well as new application development. As we have seen, this gives IT operations and architects a common platform to seamlessly extend existing datacenters into a next-generation cloud model, leveraging VMware support and a robust ecosystem of trusted cloud partners to deliver consistency in a multi-cloud world.