The myth of multi-cloud
Multi-cloud is a more well-known concept than poly-cloud. It refers to a cloud adoption model for spreading your existing and new workloads across multiple cloud providers – typically Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP). Fundamentally, multi-cloud is focused on reducing a dependency on one provider, particularly with fluctuating prices.
However, multi-cloud is not the optimal solution. Every cloud platform has excellent advanced services in the areas of databases, serverless, AI, machine learning and more. Adopting multi-cloud is fundamentally limiting because, although you can use multiple clouds at the same time, you are not optimizing which cloud you use for which service. Whilst businesses have been trying to implement multi-cloud for years, it has arguably brought very little value in comparison to sticking to a single provider due to its distinct lack of specialisation.
Enter poly-cloud.
Why poly-cloud?
Poly-cloud is an evolution from the standard multi-cloud approach. Where multi-cloud places an emphasis on the generic nature of cloud capabilities, poly-cloud is focused on the specialised nature. Poly-cloud allows businesses to choose the most beneficial cloud services for their operation based on the specific features of that cloud service, irrespective of which provider runs it. Put simply: some cloud services may have what you need, but others won’t.
There are undoubtedly some services that are better than others for certain functions. It would be easy to choose to run big data and analytics workloads on GCP, all Windows Server workloads on Azure and everything else on AWS. But this is overly simplistic and stifles efficiency. For example, AWS has been running more Microsoft workloads, and for longer, than Azure and GCP combined.
A poly-cloud strategy allows businesses to overcome this and to be more thoughtful and intentional around the way they choose services and providers. To combine multiple clouds is one thing but being selective is much more advantageous in both performance and value terms.
There are, of course, drawbacks. Poly-cloud sounds simple in its theory but is complicated to implement. Setting up poly-cloud is time consuming, and it would require an honest assessment of a businesses’ current cloud architecture to understand what is required to make specific operations work within a poly-cloud approach. Crucially, it requires expert teams that have the knowledge, skills, and experience in all three clouds.
For this blog post I spoke with James Smith. CEO of DevOpsGroup and now Managing Director EMEA for Sourced Group, James has been working with clients on poly-cloud strategies for the last few years. Speaking on the multi-cloud versus poly-cloud debate, James said: “We’ve never been a supporter of a multi-cloud approach, it’s expensive for the customer because you are always having to make compromises in terms of the solution you can create. Poly-cloud absolutely makes sense. Each of the three large hyperscalers have areas in their service portfolio that are stronger than others and the majority of our customers have this strategy.
“Having some knowledge of other cloud platforms also helps allay fears of vendor lock-in for our customers. However, the limiting factor here is in skillset. You can’t hire enough engineers that have all of the skills required at the depth required in all three cloud platforms. Hiring engineers that know one platform well is difficult; for all three platforms it’s impossible.”
Managing the skills deficit
As the Cloud Academy lead at Digital Futures we teach our cloud engineers about the benefits of multiple cloud providers and technologies so that poly-cloud can become a reality for our clients. We encourage an open mindset when it comes to what cloud service can be used for each solution.
AWS forms the bulk of our curriculum, with engineers passing their Cloud Practitioner and Solutions Architect Associate examinations within 12-weeks. Alongside this, we place a major emphasis on standard cloud technologies such as Linux, Terraform and Docker. With firm foundational cloud knowledge in place, learning Microsoft Azure and GCP then comes easily to the engineers and ensures they are ready to demonstrate their skills across multiple cloud providers.
With such a holistic understanding of the cloud, it will be these future engineers, unburdened by bias and profiting from a huge amount of curiosity, that hold the key to a successful poly-cloud approach.
Top comments (0)