AWS Cloud Enterprise Strategy Blog

Proven Practices for Succeeding with a Multicloud Strategy

MC

As an Enterprise Strategist, I’ve noticed that discussions about multicloud are often marked by confusion and contradictory advice. Some advisers warn against adopting a multicloud strategy, while others suggest that you will miss an industry-wide transformation if you don’t. There are legitimate reasons for and against multicloud strategies; success depends on balancing potential business value against complexity and risk.

Organizations typically adopt multicloud for strategic reasons. They integrate newly acquired companies that operate on different platforms, leverage specialized capabilities from different providers, or support different cloud strategies at holding-company versus operating-company levels.

But companies should avoid basing their multicloud strategies on common misconceptions about universal adoption, vendor lock-in reduction, availability improvements, and cost advantages. (For a deeper exploration of these considerations, read my previous post on developing a multicloud strategy.)

Succeeding with a multicloud approach requires cloud platforms that work seamlessly with your existing tools and future choices. You shouldn’t have to rebuild everything when adding capabilities from another CSP. Nor should you need to become an expert in every platform. (This is why AWS builds connection points directly into our multicloud services. We design our tools to simplify management across providers while maximizing the performance of workloads running on AWS.)

Based on my experience with AWS customers, I recommend these practices for success:

1. Have a Clear Strategy and Governance to Support It

Just deciding to pursue a multicloud approach is insufficient; you must also establish a strategy for delivering on your multicloud objectives, including clear governance for which workloads will go where and why. Choose evaluation criteria that will optimize workloads and their dependencies. If you leave these decisions up to individuals, the resulting uncoordinated sprawl across CSPs will likely erode any value you sought to achieve. Evaluate CSP performance regularly and use your assessment for CSP selection, workload distribution criteria, and future usage plans.

It is important to have visibility into the total number of services, applications, and components across the enterprise to support your governance strategy. To gain that visibility, create a robust tagging strategy that spans CSPs and establishes clear ownership, usage, and environment (e.g., development, QA, stage, and production). If an owner cannot be identified, the resource should be removed. This codifies governance and automates enforcement without blocking progress (“guardrails, not gates”).

Cost, operational processes, and security must be monitored and acted upon in the same manner across CSPs, with the same depth of data and transparency.

2. Do Not Spread Contiguous Workloads across CSPs

Single workflows spanning multiple CSPs introduce needless complexity, risk, and cost while complicating support, deployment, and architecture—with little value added.

Contiguous workloads often involve large volumes of data that need to be processed and analyzed together. Distributing the data across multiple CSPs can create challenges in data movement, synchronization, and consistency. And managing a contiguous workload across multiple CSPs can be complex and time-consuming. It requires dealing with different APIs, management interfaces, security models, and operational processes for each CSP. Complexity increases the likelihood of errors, adds to operational overhead, and can hinder agility and scalability.

3. Have a Longer-Term Integration Strategy

Be careful when moving data between applications in different clouds, especially when compute/applications are deployed in one CSP and data storage in another.

To decide on a location for workloads and data, you should consider your long-term needs for integrating the data and workloads with others. Will the data be needed for advanced analytics or ML beyond its current scope? Will the services be consumed broadly across other CSPs or isolated to the workloads in that CSP?

For more guidance and a decision model for deployment considerations, check out my former colleague Gregor Hohpe’s post: Multi-cloud: From Buzzword to Decision Model.

4. Use Containers Strategically

Containers are often a good idea for modern applications, and they help with many aspects of portability. They are platform-agnostic, meaning they can run on any cloud platform or infrastructure that supports containerization. This allows you to develop and package your application once and deploy it across multiple CSPs or on-premises environments without significant modifications.

But be cautious: Containers do not work in all cases (e.g., large monolithic applications), nor do they solve all the issues around portability between CSPs (especially data, policies, and security).

5. Have a Single Cloud Center of Excellence (CCoE) but Specialize Within It

As we advise many AWS customers, you should form a CCoE within your organization to provide leadership, standardization, and acceleration of your cloud journey. When it comes to multicloud, we find companies are more successful when they have a single CCoE but specialize within it for the skills, tools, and mechanisms particular to each CSP. Separate CCoEs for each CSP often lead to divergence, reengineering, and waste.

6. Make Sure Security is Always a Top Priority

Managing multiple security models increases your attack surface and creates gaps. Multicloud requires companies to deal with multiple CSP security models in areas like identity management, network security, asset management, and audit logging.

The resulting complexity makes transparency harder and increases the burden on security teams, elevating risk. Several security practices become more important: (1) shifting security left by automating and embedding it into delivery pipelines, cloud environments, and team priorities; and (2) encrypting data at rest and in transit within or between CSPs.

It’s useful to designate a single destination for security operations data (i.e., a single pane of glass). Then use each CSP’s cloud-native tools to best present the data for its environment.

7. Embrace an 80/20 Approach Over Equal Distribution

How you distribute workloads across providers affects your multicloud success. If you concentrate 80% of your investment with a primary provider and use others for specific capabilities, you can reduce cost and complexity.

An 80/20 distribution accelerates innovation by letting your teams develop expertise in your primary platform’s advanced services. You reduce training and tooling duplication. And you only need to manage one security model instead of multiple models across different providers.

When engineers master one platform, they build more efficiently, troubleshoot faster, and implement more sophisticated solutions. Companies also report better talent retention—their teams develop valuable, marketable expertise instead of thinly stretching themselves across multiple technologies.

Learning More

To learn more about the AWS services that can help simplify management and monitoring of multicloud environments, provide access to all your data wherever it is stored, and harness generative AI in multicloud environments, check out Multicloud on AWS.

Tom Godden

Tom Godden

Tom Godden is an Enterprise Strategist and Evangelist at Amazon Web Services (AWS). Prior to AWS, Tom was the Chief Information Officer for Foundation Medicine where he helped build the world's leading, FDA regulated, cancer genomics diagnostic, research, and patient outcomes platform to improve outcomes and inform next-generation precision medicine. Previously, Tom held multiple senior technology leadership roles at Wolters Kluwer in Alphen aan den Rijn Netherlands and has over 17 years in the healthcare and life sciences industry. Tom has a Bachelor’s degree from Arizona State University.