Cloud 3.0: Is a new cloud ecosystem forming right before our eyes?

It's a new year and you know what that means: new buzzwords in technology driven by the hype machine! This year (so far), the hot ticket is Web 3.0.

Web 3.0 or Cloud 3.0?

This article is not going to sing the praises of Web 3.0 or tear it down. I will let the marketing geniuses and technology pundits do that. This newsletter is for data professionals and, to that end, I want to keep the focus on data.

Despite the snarky comments about about buzzwords and the tech hype machine, I am introducing a new one in the following paragraph. I guess irony can be pretty ironic sometimes.

Hear me out:

Cloud 1.0

Cloud 1.0 was about moving compute and storage off premise. Enterprises could rid themselves of the job of managing data centers and focus on more strategic IT initiatives. Requisitioning new hardware, be it storage or compute, was now a matter of a few clicks or lines of script. This was the dawn of DevOps.

Cloud 2.0

Cloud 2.0 was about differentiation. The cloud was now a Platform™. Big cloud vendors sought to add value by offering solutions built to ease development and migration to the cloud. And by cloud, they meant their cloud.

Want a cloud-based data warehouse? Sure thing: here's AWS RedShift or Azure Synapse.

It's easy to get started. Better yet, the TCO was lower as much of the underlying infrastructure, hardware, security patching is handled for you, This was the dawn of PaaS, or platform as a service.

There is, however, a small catch. To quote the Eagles "You can check out anytime you want, but you can never leave." Once you pick a platform, forever will it dominate your destiny. You and your enterprise are now joined at the hip with the cloud vendor. Sure, you can always migrate away, but the sheer labor costs, risks, and egress charges make that rather unpalatable option.

In my experience, there were three general reactions to this from customers:

• Accept and Embrace: putting all your eggs into one basket and watching that basket closely
• Diversify and Distribute Risk: allowing each division to pick a cloud vendor or, for the more anarchy-inclined, picking a different cloud vendor for each project/application.
• Fear and Loathing: keeping everything on premise and ignoring the cloud aka "Server Hugging"

Each one of these approaches has their pros and cons.

Very often, the culture of the organization drives the decision rather than technology. Naturally, decentralized companies go for multi-cloud, while the opposite holds true for organizations with a "one standard to rule them all" approach.

As a technical seller, it was very easy to predict which customer would go in which direction.

The Fear and Loathing crowd, however, was always more interesting. Most were regressive organizations that feel that they need to control everything and that no cloud vendor could possibly understand their special unique IT requirements. A few really did have very specific requirements that no cloud vendor could satisfy.

Then there are the "chess players," who are thinking a few moves ahead and know that once they commit to one cloud vendor or another, they lose some operational flexibility. Should prices on cloud services go up or terms of service change, they know that they'd have little recourse.

These were the customers who knew that once they hand over the keys to their data services, that regaining control would not be trivial. Naturally, these were the hardest customers to convince to implement native Azure solutions. I would imagine my counterparts at AWS, Google, et. al. would say the same thing.

That is until the advent of services that ran more or less the same on multiple clouds -- an approach that I call Cloud 3.0.

Cloud 3.0

If you look at Databricks or SnowFlake, you will see something rather unique about them: they run on multiple cloud platforms more or less the same. That may not sound revolutionary, but it is.

For instance, as a CTO I can confidently state to my CEO and stakeholders that we are leveraging the power of the cloud, but keeping our options open as to which vendor we use.

Should Cloud Vendor X raise prices or do something unsavory, I could remain calm and port our solution and services over to Cloud Vendor Y or even to Cloud Vendor Z.

Sure, there'll be costs and it won't be trivial amount of effort, but it is possible. I could redeploy on another cloud without tearing down everything and building anew.

This is where I see the next wave of innovation to be: native multi-cloud. You're starting to see the beginnings of this being a wider strategy.

Just look at Azure ARC or SQL Server.

SQL Server runs natively on Azure, as a Managed Instance, as a virtual machine, and on ARC. SQL Server can also run on AWS.

If that weren't enough, SQL Server has also been containerized, meaning it can run just about anywhere.

Bottom Line

The data scientist in me thinks that it's still too early in the wave to say if this is a trend or a fluke. If you look the rise of both Databricks and Snowflake, clearly they are successful and were successful before they ran on multiple clouds.

Has their success increased since porting to multiple clouds? Yes.

Has this continued success been due to their multi-cloud strategy? Probably.

Is this the dawn of a new multi-cloud ecosystem? Yes

Are there still very real engineering obstacles to overcome? Most definitely.

Will major cloud vendors accept this model despite it not being in their best interests short term? Probably.

They will follow the money and find new ways to differentiate and entice customers to use them first, rather than exclusively.

What do you think? Are we at the dawn of a new era for cloud or is this just a phase?

Thank you

Thank you for subscribing to my newsletter.