The Risks of Multi-Cloud Outweigh the Benefits
As most companies look to embrace the public cloud, one question stirs frequent debate, “Do we go with just one public cloud provider or spread our use across multiple providers?”
As your organization evaluates this problem, it is important to understand the economics, impact of choice of services, skill base of your organization and potential complexity of a multi-cloud vs. single-provider approach. The purpose of this article is to outline some of the key attributes to help your organization make this decision.
1. Cloud Economics
Consumption economics has shown that cloud providers continually strive to offer their services at the lowest price, based on customer demand over time.
In the world of cloud, many organizations are learning that IT cost models are shifting from a CAPEX model to an OPEX-driven model. To this end, each cloud provider not only offers on-demand pricing to their clients, but they can create long-term enterprise purchasing agreements to drive greater value to their clients. Examples include:
- An existing relationship with Microsoft can bring additional financial benefits as clients can bundle existing licensing agreements and volume discounts with Azure spend.
- VMware can offer AWS Services credits in a larger Enterprise Licensing Agreement in order to motivate customers to adopt VMware Cloud on AWS.
- AWS can take advantage of its Enterprise Discount Program to not only get lower pricing based on volume commitments, but they can also use the program to consume in the AWS Marketplace.
Key Takeaway: The biggest economic benefit from cloud adoption comes when an organization can commit to volume consumption over long periods of time, typically 3 years. With most organizations, especially as they are scaling their initial use of a public cloud platform, it is difficult to amass enough spend across multiple vendors. Therefore, aligning themselves with one provider to maximize near-term discounts and special pricing is most beneficial.
2. Using Best-of-Breed Cloud Services
It’s hard to say one public cloud is better than another; they each have unique and compelling services. By maintaining multiple cloud offerings, organizations can leverage the best services within each cloud.
For example, IaaS virtual machine services are fairly consistent across providers, however advanced Machine Learning, Artificial Intelligence, Database, Application Services, IoT, and Automation services can be particularly differentiated, with each of the major cloud providers maintaining an edge in one or more of these categories. These advantages ebb and flow between providers on an almost monthly basis.
In the enterprise, applications rarely live in isolation. They are constantly integrated into other applications, databases, security and monitoring services. To this end, leveraging services from different cloud providers within a single application can add complexity – not just from a development perspective, but even more so from an operational one.
To this end, organizations must take into account their long-term application strategy when thinking about leveraging multiple-clouds as this will have an impact on which services they leverage and how much IaaS, PaaS, FaaS, or SaaS they will consume.
Key Takeaway: Our recommendation is to think long-term and not let the addition of a specific feature set cause you to increase complexity by introducing multiple clouds into your environment – especially within a single application. Focus on building best-of-breed tooling, processes, and standards when building applications and infrastructure. Additionally, don’t forget that many teams still have to maintain existing applications and infrastructure, so adding not just one, but multiple cloud providers may stretch them extremely thin.
3. Impact on Skills, Training, and Hiring
The US talent pool for cloud engineers shows a very clear trend toward the majority of individuals focusing on one major cloud instead of several cloud providers. The thought is that rather than possess mediocre skills in several cloud platforms, individuals would rather go deep in a single cloud provider in order to differentiate themselves. There are hundreds of services within each of the major cloud providers, making it difficult to stay on top of everything in that cloud, let alone multiple cloud landscapes.
Practitioners tend to develop skills within a cloud provider as well as focus on an area of emphasis within that cloud. The certification paths offered by AWS, for example, are suggestive of this pattern. Individuals need to learn a base set of skills across a wide variety of services for the associate tier of exams. Then practitioners progress to the professional exams, honing their skills further. There are also specialist exams in order to test individuals choosing to progress to the highest level in specialized areas such as networking, big data, and security.
Key Takeaway: For most organizations, we recommend seeking deep technical and specialist talent within a single public cloud in order to maximize efficiency and engineering depth.
4. Operationalizing a Single Cloud Provider
In addition to the challenges around hiring and retaining talent, consider the processes, tools, and training required to operate applications and infrastructure once they have been migrated to the public cloud. This is not an easy undertaking, and often requires not only operational teams, but also deep integration with enterprise security, DevOps, and service management teams just to name a few.
Key Takeaway: Build the right enterprise guardrails, security, and operational controls and get those right in one platform ratherthan try to duplicate them across multiple cloud providers.
Concentrate First on Getting One Cloud Provider Right
We recommend most organizations focus ononboarding and operationalizing a single cloud provider before attempting a multi-cloud strategy. Doing so will create more efficiencies than trying togain economic leverage across vendors.
While there are benefits to a multi-cloud strategy, the potential risks outweigh the gains. In particular, we recommend that clients continually evaluate themselves against a single-cloud strategy focusing on the following criteria:
- Building an enterprise foundational cloud including security, governance, account structure, shared services, chargeback, and complete networking.
- Establishing an enterprise cloud operating model in order to address changes, Day 2 operations, incidents, naming and tagging, and automation.
- Evaluating tools and services that make the future adoption of a multi-cloud environment a possibility over time. This includes tools like infrastructure-as-code, configuration management and image management (e.g., Terraform, Puppet, Chef.)
- Creating a team of cloud architects and engineers with defined career paths and certification opportunities, armed with the necessary skills to support both intermediate and advanced core services in a single cloud.
- Defining a cloud strategy in which various business units have shared use cases for service offerings and innovation, which can, in turn, be rationalized by the cloud team.
- Opening a Cloud Center of Excellence (CCoE) with the mission of advancing the business goals as they pertain to the previous five points, and with the intent to advance and share knowledge about the cloud throughout the organization.
- Defining and requiring teams to use a standardized tool set, leveraging them for infrastructure and application deployment through the use of collaborative Dojo’s where you help onboard new applications into your cloud environment.