Strategy 1: Create a Unified Infrastructure Control Plane
A fundamental challenge when it comes to managing infrastructure across a hybrid cloud is that every cloud—public or private—is unique. This strategy digs deeper into the challenges of hybrid cloud operations, suggests a plan of action for addressing those challenges, and looks at the landscape of possible solutions.
Hybrid Cloud Infrastructure Management Challenges
Although things may be conceptually similar between different cloud environments, at a practical level there’s no standardization. Knowing what compute instance or what storage type to use for a workload in one cloud doesn’t translate easily to another environment.
The fact that each cloud essentially operates as a silo causes challenges to multiply.
Management Interfaces Differ from One Cloud to the Next
The tools for selecting, deploying, and configuring infrastructure services are different from one cloud to the next, and expertise in one cloud doesn’t mean you’ll be immediately efficient in another. Given the rapid growth in infrastructure and other services in each of the major public clouds over the past few years, just keeping up to date with the capabilities and best practices for a single cloud platform has become a challenge.
Siloed Cloud Management Teams
Siloed cloud management is almost certainly why most enterprises have separate teams dedicated to each cloud platform they use, as illustrated in Figure 1-1.
Figure 1-1. A typical hybrid cloud consists of operational silos, with separate teams responsible for each cloud.
Lack of Integration
There’s a surprising lack of integration between private and public clouds and across different public clouds. Public cloud providers mostly expect you to use their integration tools—and take widely different approaches to private-cloud integration.
There is still very limited interoperability between clouds or integration across clouds, since it’s to a public cloud’s benefit to lock you in. Many public clouds make it cheaper to move data in than out, and the hybrid cloud tools they offer are more about colonizing your datacenters with their technology than helping you achieve broader interoperability or integration outside their ecosystem.
These operational silos—although they may seem a necessity given current cloud realities—are an impediment to hybrid cloud success. Siloed teams make collaboration more difficult and ultimately slow progress.
Plan of Action: Unify Infrastructure Management
As a result of these challenges, most enterprises have an approach to hybrid cloud management that has more to do with trying to get the most from each cloud than it does with getting clouds to work together seamlessly. What’s needed to fix this problem is a unified control plane that gives you the ability to monitor, manage, and orchestrate across all environments with a single set of tools. This is the only way to operate at the highest level of maturity and achieve the full benefits of hybrid cloud.
The following approach can help you achieve the level of integration among clouds that you’ve been lacking:
-
Choose a single control plane that you can apply everywhere, abstracting the differences between different environments. (Some available options are discussed shortly.)
-
Modernize your datacenters to use that control plane as broadly as possible.
-
Choose public clouds where you can use that control plane.
This strategy was the main focus of my report Designing and Building a Hybrid Cloud (O’Reilly 2018), where I referred to this concept as a cloud operating framework rather than a control plane, but the advice remains the same.
Identify your Goals
Creating a coherent, top-down plan results in a more unified and efficient hybrid cloud environment. Before choosing the best unified management approach, assess your environment and think about what your goals and priorities are. It’s likely your goals include some or all of the following:
- Eliminate silos
- Eliminating operational silos will make your operations more efficient, but it’s important to identify the highest priorities across your company’s operations (or the operations that are under your control).
- Unify management
- You may need to unify management across all private and public clouds, or you may have just one or two high-priority cases in mind.
- Standardize tools
- Standardizing tools across environments increases flexibility, decreases the need for specialized skills, and offers a host of other benefits—even if your organizational structure remains siloed.
- Increase pace of innovation
- Simplifying management and enabling self-service can help developers and DevOps teams deliver products faster.
- Increase cloud security
- See Strategy 4 for more on multicloud security.
- Reduce cloud costs
- Maintaining separate operations teams for each cloud is expensive and results in a lot of redundancy. (See Strategy 5 for more on cost governance.)
Choose the Right Solution for Your Needs
There are two general classes of solution for unifying the control plane across private and public clouds:
- Unified cloud management
- Provides a layer of abstraction on top of existing cloud interfaces.
- Unified platform
- Integrates closely with the underlying infrastructure, providing the same interfaces and services everywhere.
The best option for you will depend on your requirements. Here are some possible solutions in both categories:
- Unified cloud management
-
- Hybrid cloud management
- A wide variety of hybrid cloud management solutions have emerged in recent years. The Forrester Wave™: Hybrid Cloud Management, Q4 2020 report compares nine available solutions based on a broad range of criteria, including provisioning, automation, monitoring, and orchestration.
- Configuration management and orchestration tools
- There are a number of well-known tools in this class that are popular with DevOps teams. These tools may already be part of your infrastructure stack. Although some of these products operate both on-premises and in public clouds, they can require significant expertise to configure and use in each cloud. As a class, these tools may be facing significant headwinds in the DevOps domain due to the growing popularity of Kubernetes.
- Unified platform
-
- Kubernetes
Kubernetes is a platform for automating and managing the execution of containerized applications, particularly cloud native applications that use a microservices architecture. Kubernetes by itself is not a unified platform, but Kubernetes services are now available in all major public clouds, or you can deploy your preferred Kubernetes software on the cloud(s) of your choice.
Therefore, you can use Kubernetes to provide essentially the same environment everywhere and run containerized applications in the location(s) of your choice without modification.
Although the control plane may be the same or at least similar for each cloud, Kubernetes by itself does not include the ability to manage multiple clusters across different clouds from a single control plane.
If your operations are focused entirely on containers, Kubernetes may be a good choice. However, if you need to support both VMs and containers, or your operations need a control plane with a greater level of abstraction, you will likely want a solution that incorporates Kubernetes but isn’t limited to Kubernetes.
- Platform vendor solutions
A number of prominent vendors in the IT space offer public cloud solutions that enable their platforms to run also in public clouds, creating a single unified platform spanning private and public clouds. Prominent examples are VMware and Nutanix.
Look for the ability to monitor and manage private and public cloud environments from a single control plane and to facilitate extending operations, bursting, and application migration among all supported environments.
- Hybrid solutions from public cloud vendors
Solutions such as AWS Outposts and Azure Stack create hybrid clouds by extending public cloud capabilities to your datacenters. These solutions may be appropriate for hybrid clouds that connect your on-premises datacenters to a single public cloud in situations where you aren’t worried about vendor lock-in.
If what you want is a seamless operating environment that encompasses both private and public clouds, you’ll likely be best served by one of the unified platform options. Once you’ve identified a few candidates based on your requirements, the final decision may depend on which solution offers the most compelling roadmap and vision for your hybrid cloud.
Summary
Current hybrid cloud operations suffer from siloed private and public cloud environments with separate management tools and dedicated management teams for each environment. As a result, management and integration capabilities across clouds are often minimal. A smart strategy is to choose a solution that provides a unified control plane, either through cloud management software or via a single platform available on premises and in multiple public clouds.
- Key takeaways
-
-
Unifying the control plane for your hybrid cloud will allow it to operate as a more cohesive entity, eliminating the need for siloed teams.
-
Your decision should take into account your specific goals and the set of capabilities you require.
-
Unified cloud management provides a control layer on top of existing clouds, abstracting their capabilities so that everything can be treated the same.
-
Unified platforms integrate with underlying cloud infrastructure, providing the same set of interfaces and services everywhere for uniform operations with guaranteed compatibility.
-