A new strategy for multicloud: Go deep, one at a time

October 25, 2019

A blog post Mike Kavis, managing director and chief cloud architect, Deloitte Consulting LLP

Many companies have a fear of getting locked in to any one cloud vendor, so they are going all-in on a multicloud strategy. In the process, they take one of two different approaches: Either they are choosing the best cloud for the specific workload type—one for analytics, one for enterprise grade processes, etc.—or they are insisting that every workload must be cloud agnostic. The later approach tends to burden organizations as they get bogged down chasing both technology and operations solutions.

There are major challenges with the cloud-agnostic multicloud approach. The architecture and the operations tooling and processes to run the same code on any cloud increases the complexity of building and running workloads. In addition, it prevents organizations from using the high value services of cloud providers, thus minimizing developer agility and time to market. Since developers have to build a lot of the underlying technology they would get out of the box as a service from the cloud providers, the effort it takes to build, maintain, and operate cloud agnostic solutions is substantially greater than if they leverage the cloud providers’ APIs. This often leads to cloud agnostic architectures that use the cloud as nothing more than a commoditized data center. It’s not a strategy that nets the highest value of cloud.

Reap the real benefits of cloud

The real benefits of cloud are lower costs, agility, and access to innovation. With cloud, developers have the tools to create new solutions quickly. They can focus on higher level services that manage high availability, failover, elasticity, patching, and updates automatically. Developers can then focus less time on IT plumbing and more time working with their customers, putting them in a much better position to focus on business functionality.

Say your company wants to implement blockchain. Traditionally, you would have to invest several months evaluating many blockchain technology components just to figure out how to use the underlying technologies. Only then could you begin to focus on the business problems you are trying to solve. I call this “going dark on the business” while you focus on IT plumbing. Instead of figuring out how blockchain works and implementing all of the underlying hardware, middleware, and software yourself, you might choose AWS’s managed blockchain services—it runs, it scales, and it saves your IT department up front time and energy and reduces the operational overhead required from homegrown solutions. If you’re taking a cloud agnostic approach, it could take six to eight months to stand up your own implementation. By leveraging a cloud service, you could be prototyping in weeks or even days.

While being cloud agnostic takes away many benefits of cloud, it is understandable that organizations don’t want to put all their eggs in one basket. They believe it is faster and easier to change direction if needed by spreading across multi-providers. But there’s a tradeoff: Time, cost, and complexity may stand in the way of the results they had hoped to achieve and every second spent on trying to remain agnostic is time not spent working on customer requests

Transforming an organization from a datacenter mindset to a cloud mindset can be extremely challenging. But going from a datacenter mindset to multicloud mindset right out of gate has tremendous risks due to the lack of skills across all cloud providers, lack of cloud operations maturity, and lack of cloud native architecture and development experience.

Let’s consider another approach: Why not get proficient at one cloud first, then extend to other cloud providers once the organization has built the necessary cloud muscle to move forward? Cloud providers innovate incredibly fast and it is hard enough to keep pace on the capabilities of one cloud provider, not to mention multiple providers. Finding good cloud talent is a challenge, but it is nearly impossible find talent that has deep expertise on multiple clouds. Growing your talent on one provider is a better way to increase cloud maturity in your organization. Once the staff is proficient on one cloud provider, then they are better suited to tackle the next cloud provider.

Mitigate the lock-in

Getting the first application to the cloud is usually a very transformative event for an organization that can have major impacts on the technology, the development, and deployment processes and the operations of that application. Insisting that the first application can run on all major cloud providers only increases the complexity and timelines for getting that first application into production. Lock-in is a valid concern, but at what cost? Try mitigating lock-in by taking a different approach:

1. On the back end. Choose to be cloud agnostic at the operating system level. Don’t use the cloud provider’s services when it comes to configuration, patching and installation; use cloud agnostic solutions and OS images.
2. On the front end. Don’t use the cloud providers visualization tools. Instead use third party solutions that support multiple cloud providers like Tableau or Qlikview for example.
3. In between. Go very deep into the cloud provider’s stack: Use fully managed services like database as a service, blockchain as a service, and streaming services, etc. to maximize agility and reduced operational costs. It’s the best way to extract value from the cloud and drive more business value for your organization and customers.

With this approach, you can put new applications together very quickly. Then, if you need to go to another platform, you don’t have to replatform the entire application. 

Some companies are three years or more into their cloud journey and have extracted very little value out of their cloud implementations because they have accomplished nothing more than establishing virtual datacenters on multiple cloud providers. They are still doing much of the same work they have always done in the datacenter, but now with different tools and with less expertise. However, many companies who have selected one cloud initially and gone deep in that cloud provider’s stack are often the ones who have been most successful. They’ve put new or replatformed apps in the cloud, learned from them, and then continue to improve and mature. These organizations recognize that while a multicloud strategy is in their future; for now, they have chosen to lean on a preferred cloud provider as opposed to being cloud agnostic and are reaping the benefits of the cloud by leveraging the cloud provider’s higher-level services where the true value of the cloud lies.