7 vs. Wild – How to make sense of a Jungle of Clouds

Aug 17, 2022

AWS, Google Compute Platform, Microsoft Azure, IBM Cloud, Oracle Cloud, Alibaba Cloud or Salesforce Cloud – There are dozens of cloud providers on the market and you might be wondering how to make use of them.

In this article we will discuss the state of competition between these cloud providers and give an indication of what criteria you should look at in developing your own cloud strategy. As you can imagine our focus will be on the data, analytics and machine learning focussed tools, but most thoughts apply to traditional workloads as well.

Welcome to the Jungle: State of Cloud Competition

The first question we typically get asked is: which cloud is better in terms of features, cost, data security, usability and other categories. And while that is a legitimate question to ask, it is not the most important one to ask in 2022. This year we see a very healthy, competitive cloud market and as consumers we greatly benefit from that fact. Costs have largely become a non factor. The “big three” AWS, Azure and GCP continue to drive down prices. Almost every other year one of these players takes over the lead and pushes down the prices further. Being less feature complete, we sometimes see even lower prices on the smaller cloud providers. As you can see in the table below, prices for raw compute are largely within 15-20% of each other. We see a similar pattern on the storage side of things.

Source: https://www.simform.com/blog/compute-pricing-comparison-aws-azure-googlecloud/

What about feature completeness then? Once again, it hardly matters. Someone has a great idea or releases a managed version of a great new open source tool first, but is bound to be copied by the others in less than 12 months. That kind of feature gap is not enough to make a choice for a cloud provider. Take care when picking one of the smaller providers, but for the big three it should not be your guiding principle.

Source: https://medium.com/@vineetjaiswal/introduction-comparison-of-mlops-platforms-aws-sagemaker-azure-machine-learning-gcp-vertex-ai-9c1153399c8e

Now, what about momentum and market share? This is indeed a key factor. As the graphic below shows, the market is split into a healthy set of providers, but AWS and Azure have taken a decisive lead. You will notice and feel this fact around every corner. Their ecosystem is more mature, more supported, more battle tested. Found an interesting new tool that you want to run? The odds of it running out of the box on AWS are very high. Looking to use terraform to manage your infrastructure? The terraform providers of the three big cloud providers are in tip top shape, while the others might not be.

Source: https://www.wpoven.com/blog/cloud-market-share/

Building a shelter: Choose your cloud

So, where does this leave us? What can help us make a decision apart from momentum and market share? First you must figure out if cloud makes sense for you, at all. In our space of machine learning, it often does: Machine Learning training and data transformations are rather bursty by design and we greatly benefit from the elasticity of the cloud. Need 100 Graphics cards for a few hours to quickly train a model, then discard them? This is great on the cloud and would never be feasible with physical hardware. Most traditional workloads benefit from this flexibility as well, but typically only when you re-engineer rather than simply lift and shift.

Next consider your context: Got a running Active Directory? Or a set of engineers that already has certifications in cloud X? Got applications that lend themselves to one of the providers? What can help in the transition? In general, there are many questions here and no very obvious answers. Feel free to send me a message as well, I’m always up for talking about this subject.

At 26% Europe has the lowest cloud adoption rate of all developed countries

My final subject is the topic of multi cloud. At 26% Europe has the lowest cloud adoption rate of all developed countries. This is further exacerbated by a somewhat irrational fear of committing to a single provider. The most cited arguments are worries to be locked into a single vendor and the ability to leverage optimal prices across cloud providers. But, let’s face it: for 99% of businesses these benefits will never come to pass or be overshadowed by downsides. Running multiple clouds is a significant overhead to IT organization and requires dedicated, excellent skills on all involved clouds. So my recommendation is to start on one cloud, really get it right and go from there. Build excellent teams that quickly demonstrate value. Use the chance to apply modern methodologies such as Infrastructure as Code, FinOps and GitOps.

Surviving the island: Don’t go alone

It goes without saying that every business is different. Don’t take any of the insights shared and apply them without reflecting them in your own context. And feel free to break rules as well. When creating momentum for going cloud in the first place a case could be made that things should be done in a more pragmatic, proof of concept kind of way. Last but not least: if you want to discuss how to go cloud yourself and need a partner on that journey – don’t hesitate to get in touch with us. Over the last years we have created more than 30 cloud accounts for our customers and are happy to do the same with you.

And yes, if you have been wondering what this quirky 7 vs. Wild humor is – we love the show and are very excited for season 2!

Ferdinand von den Eichen

Weitere Blogs

Jan 26, 2023

AI: Build it or Buy It — 6 Reasons for Each Approach

Jun 15, 2021

Building the AI Factory of 2021 and beyond— A Journey

contact us

Realize your AI plans now

We look forward to getting to know you with a no-obligations conversation. Contact us now and we will get back to you immediately.

Kineo.ai team
Thank you so much for
Your enquiry

We'll get back to you as soon as possible.  
‍In the meantime, have a look at the other pages.

Oops! Something went wrong while submitting the form.