I read a great article about the exit from ready-to-use cloud platforms. I can sign every single word. As far as I see, they switched their infrastructure from ready cloud servers to custom cloud servers. So they’ve been moved out of using some ready cloud services and their goofy features, which is great. I hope many will join or follow them. Let’s call it “CLEXIT”.
We started up ARSUITE about 7 years ago, and in ARSUITE, we never use any ready cloud platforms or services because we believe they never provide any enough resources depending on their promises and charges. I like and believe in SaaS and PaaS, these are different topics, but renting some insufficient storage, functions, and databases under the name of cloud computing is very expensive and never wise.
I’d like to revisit the same questions in a different order and share them with you, with different answers depending on our experience. Please notice that we are talking about running some advanced services and APIs on some cloud servers. Here, we are not discussing subscribing to or using some ready-to-use SaaS or PaaS services or products. We will review ready-to-use cloud platforms.
Why didn’t you just optimize your cloud bill instead?
We can’t optimize the ready cloud platform invoices because we don’t fully understand the mathematics of the ready cloud platform billings. I don’t think the platform owners or employees even really understand that math. Can anyone explain to me in detail how AWS, Azure or other billing math works? “I can’t”
Won’t your hardware savings be swallowed by a bigger team payroll? Don’t you need a world-class team of super engineers to do this?
Ready or Custom cloud server running and maintaining requires the same team composition and member count because ready cloud services don’t offer everything at once or all in one. In every circumstance, we have to operate and maintain our servers and APIs ourselves.
They provide us with some of the Storage, DevOps Pipeline, Databases and Functions, and we have to develop and manage them all. Yes, we can order more support or wider features, but if we do, our bill will be incredibly high. Is it logical to increase the bill for the cloud services with the same team composition? Still, the cost of the team will not decrease, complete opposite it will increase.
Imagine you have a couple of DevOps engineers. Dockers, Kubernetes, and OS must be in their fundamental skill set. However, you need some engineer with some extra skillset when you decide to use some Xxx cloud platform. They have to know how to use or run Xxx Cloud services. As you know more skills require a larger salary.
What about setting up and stacking servers and pulling network cables? Who does that?
Building a private cloud infrastructure doesn’t mean we have to build or invest in our own data centers. There are tons of ready-to-use data centers that we can use. Don’t worry, AWS, Azure and Google Cloud use them too. Check out the link, I assure you, you will be very surprised. A data center sample. You can find many other DC providers too.
We just need to buy/rent some dedicated servers and run them in someone’s data center. I can assure you that this is more resource-efficient and cost-effective. So this strategy is more efficient in terms of server resources and more affordable in terms of costs.
Many storytellers lie or mislead you about your needs or requirements. For example, they tell us that using ready cloud platforms is well-suited or best for you if you have a small business. Is it true? Ultimately, they need to sell something to survive.
But 95% of the expertise you need to run your own hardware is the same as what you need to run the ready cloud platforms. Ultimately, we have to repeat that for every option, whether custom or ready cloud platform, we need team members and similar expertise to manage them. At the latest when you reach scale with millions of users, your bills and calculations will no longer change or increase dramatically in the private cloud server option.
What about reliability? Won’t the cloud do it for you? What about international performance? Isn’t the cloud faster?
The requirements for reliability, scalability, performance and CDN are the same. With both options, you need to maintain at least 2 replica servers in different locations. If you order some ready cloud services, the platform will ask you the same questions. For example, do you need a replica, a CDN, a backup, etc.? Please test it.
Misconfiguration and Misarchitecture
Last year I read in one of my friend X’s threads how they lost $200,000 Google Cloud credits. They had misconfigured the Google Cloud service and lost $200,000 in just one month. Yes, using ready cloud platform services can cause such idiotic and extreme problems. I believe that in most cases, we need more experienced and expensive people to manage and operate ready cloud platform services and features.
For example, we lost $65,000 in ARSUITE custom cloud server costs over 6 years because we made some poor design decisions. These decisions incurred some costs but did not affect our performance or capacity. We just used some extra servers that we never needed.
You lost $200,000 in 1 month, someone lost $65,000 in 6 years, please do the math. When you build a custom infrastructure, you need to know everything down to the smallest detail, and that reduces your margin for error. Because you will be more meticulous and you have to learn more.
How much do you spend on service contracts and license fees?
Zero. First, we never use other frameworks or tools at the heart of our platform, especially licensed ones. However, you may need some licenses and have to buy them in both options. You never run away from that.
So far we have talked about visible facts. We can guarantee you ready cloud platform services are 10x more expensive than custom cloud installations around the server maintenance perspective.
Invisible or Hidden Facts
Let us talk about Godzilla or the unspoken. We prefer Godzilla over the elephant because it’s not a simple mistake, it’s a destroyer like some huge mistakes.
What if you had used the cloud service functions or databases? “like Lambda or DynamoDB, etc.”
Now you are screwed, because if you use one or multiple of the cloud functions, you are their slave or money tree. From this point on, you can no longer change your infected APIs or transfer them to other platforms or your custom platforms.
Eventually, you can move out, but what cost? We can say that your all APIs have been infected by their cloud functions or databases. Now, you know, you are an addict and needy. If you want to move out, you need to remove functions or databases and rewrite all the infected code. You will feel like you’re in rehab, you are trying to remove all drugs in your system. I hope, now you have seen the destroyer with your own eyes. I do not think we need any further explanation. This will probably cost you 100x more. Please do the math.
If Ready Cloud Platforms are so expensive, why did we go there in the first place?
Because in some cases we bought the marketing pitch and in other cases, we were trying to protect our comfort zone. Tell your boss who is responsible for the problems you have had or who caused them. You or an Xxx cloud platform? Imagine which one feels easier.
We can’t explain everything within math, technology or engineering, we need to know and understand the psychology or motivation behind people’s responses, actions or behaviours.
The most painless happiness comes after zero responsibility and a greater safety zone.
Alper Akalın
Weak people become drug addicts when they are seeking happiness in the wrong place. And, as you know, addicted people forget and move away from everything when they get their drugs. Xxx cloud platform, Yyy technology, or Zzz framework can create a delusion that feels like happiness: “zero responsibility + stronger safe zone.” I believe this rule determines most people’s, engineers, or companies’ actions.
From this perspective, we have to accept that most developers just find it easier to submit to someone’s power monetarily and psychologically. In my perspective, they are escapers like drug addicts. Escaping from freedom into the warm embrace of some Big Brothers feels easier and safer. “Happiness, remember”. Dare to face problems or develop something independent needs a strong heart and a steel will.
I am proud of my team and the ARSUITE platform because we never fell into the ready cloud platform traps or delusions. At the beginning of ARSUITE in 2017, we designed and built our own private cloud platform. We never use ready-to-use cloud platform services, and we never paid a few million; in last six years we paid at most 300,000 dollars. The magic is here ARSUITE is an advanced cloud-native 3D, AR, VR, XR design, development and rendering platform. I believe you get the picture “3D”.
We’ll talk about our infrastructure in another blog post. We’re just trying to tell you that you don’t need over-glorified, ready-to-use cloud platforms.
Here is the David’s The Big Cloud Exit FAQ article. Thank you David. I believe that the title should not be like “…Cloud Exit”. Most developers or techies will get it wrong, it’s in their nature. They will think it is a real and extensive exit. Then, let us call it “CLEXIT”.
Stay tune,
Alp