Cloud of Freedom

Cloud of Serfdom or Cloud of Freedom

At its core, the cloud is a fantastic innovation — I do not think many of us will long for the days when it required days, weeks, or even months to provision a new server.  Programmable, instantly accessible infrastructure comes with so much flexibility. It allows new deployment and management approaches we could not even have imagined in the “old world.”

In the early days of the cloud, vendors very much embraced it, and focused on the new commodity the cloud was to become:

Electricity, mobile networks, and Internet access are wonderful commodities driving so much innovation purely because they are commodities. These commodities are accessible to everyone and (in general) have low switching costs. This is due, to a large extent, to the fact they are based on commonly agreed official or defacto standards.

The problem with a commodity though is, while it is of value for consumers and society at large, it is not so great for the company providing it, as competition inevitably reduces profitability and is not rewarded well by the stock market either.  Just compare AT&T’s P/E of around seven to Amazon’s, which is close to 90.

While initially, clouds were providing largely commoditized “Infrastructure as a Service,” they have since all moved, pushing proprietary services which exist only on their cloud.  Even more, they have invested in educational content which calls using such proprietary services as “best practice”.  If you’re an Amazon Certified Architect, you need to know you should be running Amazon Aurora, DynamoDB, or RedShift rather than using IaaS and rolling out your database on Kubernetes.

If you should follow this path of Serfdom, you will find yourself in a situation where your vendor has all the power.  It is not only pricing power but the power to shut your business down, because of international politics or because of the influence of cancel culture zealots.

Students of history may remember what happened when a young company, which was to become known as Oracle, took on IBM.

Larry Ellison

In the early days, it was saving customers from hardware vendor lock-in which came from the de-facto monopoly of Big Blue (IBM). As enough customers were sufficiently “saved” (were deep enough with Oracle technology they were not moving anywhere), Oracle gradually moved on to become the hardball company it is known for. There is a saying in the industry “Oracle does not have customers, Oracle has hostages.”

If you choose to lock yourself in, with highly differentiated proprietary technologies, you should expect the same, maybe not tomorrow or next year, but it is inevitable. As an example, you can check for yourself the difference between EC2, RDS, and Aurora and how it only gets larger as time goes on and new instance types become available.

There is one more point to consider — over the last decade, cloud spending (and so revenues for cloud providers) was growing in the double digits because so many workloads were moving to the cloud.  As we complete this move though, the growth cloud vendors seek will need to come from charging existing customers more.

This will especially be a problem for larger companies who have substantially locked in with the cloud of their choice and are perceived to pay the ransom, as they have no other choice. Early-stage companies and startups will continue to be pampered including receiving generous free credits to get started. Hey, drug lords already figured it out – to build the most loyal customer base you better give the first dose of heroin for free.

Cloud of Freedom

There is another way though, and like embracing open source in the late 90s, it requires taking the road less traveled and putting in more elbow grease, but it comes with substantial cost savings and most importantly freedom of controlling your infrastructure, your data, and being in a position to pick and choose cloud providers to your liking.

The idea is simple — use a Cloud Vendor as a commodity infrastructure provider and use Open Source Software for higher levels of the stack.

You may rightly point out that open source solutions are not as polished as a well-integrated set of services you get from AWS and other cloud vendors. It is true, of course. If you have been in the industry for a while it might remind you of a situation with Linux in the late nineties.

I remember in those years, I chose Linux as the operating system for the startup I was building. My friends working for serious companies were laughing at me — Linux was so immature in terms of usability or performance compared to Solaris, HPUX, AIX, and Windows NT that it was almost a toy.  At one point you could not even have files more than 2GB in size.

Yet we all know how that story ends — the industry came together and improved Linux to the point where it is a very mature, scalable, and high-performance operating system that underpins most of the Internet.

One important contributing factor which pushed many companies, especially startups, to embrace open source was the Dot Com Crash of 2001.  Linux, Apache, MySQL, and PHP (LAMP) did not magically become much better than Microsoft or Java stack.  Many companies just could not afford them anymore, and in the end had to use open source technologies, and improve them along the way.

Where we use to have Linux, in the cloud age we need more than that, and we’re getting it with Kubernetes and the cloud native ecosystem.

https://landscape.cncf.io/

The cloud native ecosystem is still young compared to the cloud, but it’s getting a lot of investment and growing rapidly.  You just need to attend KubeCon/CloudnativeCon to see this ecosystem is going places!   You see hundreds of companies innovating in their own way — it is messy at times, and there are often multiple options and multiple opinions on how to approach a specific problem. We know other long-term approaches tend to provide a better outcome, for example, democracy, with all of its inefficiencies, tends to win against dictatorship.

Yes, of course, AWS, GCP, Azure, and other clouds also have wonderful partnership ecosystems, but it is different. It is never about working on the best solution to the problem or achieving the best outcome for the customer. Instead, it is about one dominating player setting the rules of the game with their own commercial interests and others choosing to play, or not.

Path to the Cloud of Freedom

So, let’s say I have convinced you to embrace open source, and using the cloud as a commodity is a better path, even though it might not be easy. How should you embrace it?

Let’s be real — in its current state, even though fantastic progress is made every year, the cloud native ecosystem may not have robust solutions for everything major clouds have to offer.  If you’ve been on the proprietary cloud bandwagon, it is likely the team has more knowledge of proprietary cloud approaches, and as such, would need to go out of their comfort zone to embrace cloud native solutions. The practices to follow are basically the same which helped open source to dominate in the data center:

• Strategically embrace open source solutions in the cloud, create incentives to overcome friction, and create situations where only what is really not feasible with open source uses proprietary cloud services. Focus on new applications first — it is much easier to build new software on a new stack than retrofit the old one.
• Support open source developers, and vendors doing development.  Software does not build itself and we collectively need to fund the development and maintenance of software if we want robust, scalable, secure software to exist.
• Participate in the open source ecosystem yourself.  The best way to make sure open source software meets your needs is to make it so. Even if you can’t contribute code, contribute bugs, ideas, documentation, and content.

As often happens with open source, we have proprietary vendors figuring out what exactly works in the market and educating the masses about new approaches, now is the time for the open source community to build on it and bring freedom into the cloud.  I’m sure 10 years from now, the cloud will look nothing like it is today!