Nebulaworks Insight Content Card Background - Denys nevozhai la interchange

Nebulaworks Insight Content Card Background - Denys nevozhai la interchange

Kubernetes is not a DevOps Cure-all

February 17, 2018 Chris Ciborowski

Buying a tool does not equal transformation, and that applies to Kubernetes.

Recent Updates

Tony Stewart

Have you ever felt a need to climb to a high place and proclaim something? I am guessing that you have…it’s human nature. Especially when you are passionate about it. I have something on my mind that I need to get out there. It’s been eating away at me for years (decades really). Ready? Kubernetes is not a DevOps cure-all, and

Buying a tool does not equal transformation!

There. I said it. They don’t make your organization better or high performing. And to think they do is outright foolish. You may be thinking “well, no $hit Chris. Tell us something we don’t know.” OK! Stop reading because you agree with me. But for the throngs of others who routinely put a tool, like buying Kubernetes (or commercial Linux distros) in front of adopting new methods and optimizing processes to transform, you should read on. We’ll explore this in the context of something I’m passionate about; Motorsports. Stay with me, it’s a good read.

See, I am a car guy. Ever since I was a little tike there was something about the automobile (anything mechanical that moves, really) that intrigued me. The fact that you were in control of something that could drive around - in some cases VERY fast - was captivating. And that is the type of driving I became obsessed with. Indy. Formula 1. Yep, glued to the television to watch. And I made a promise to myself that if I ever had the chance I would drive on a racetrack.

Fast forward.

As an adult I was able to attend a high-performance driving school. YES! I’ve been watching racing forever and knew exactly what I needed to do. Get me in the BMW M3 over there and you’ll see a display of Michael Schumacher-like skills. Why? Because M3. Over 400 ponies under the hood, shod with fresh Continental tires, tied together by a dual clutch transmission. 0-60, 4 seconds. Just give me the keys already!

Truth be told, I’m not slow by any stretch of the imagination. After securing the title for the handling course in the M3 at the BMW driving school, I asked an instructor how my time ranked outside of our class. He said, and I quote: “You’re consistent and pretty quick. Well done. But our instructors run those times in the rain.” Whaaatt? How could that be? 40% less grip and as fast? I had an M3! What did they have I didn’t? It had to be a different car with more power. That’s it!

So, when I had the chance to acquire a high-performance car I did just that. A 507 HP BMW-M monster. V10 engine. Unstoppable! Faster lap times ensued. Still not good enough, I upgraded again. 560 HP. Oh yeah! Nobody is catching me. And frankly, not many did at our track days. But I yearned for better times. Wanting to go even faster ‘round the track I decided to get a sports car. Smaller == faster, especially considering my “driving skills.”

Boy, was I wrong.

First outing in the sports car at the track I proceeded to get my butt handed to me. My lap times sucked. Couldn’t even get close to people I was running down and passing in my old cars. Why? I had one of the best sports cars on the planet. Here’s why:

A tool is no substitution for practice and expertise

I had learned a VERY tough lesson. Just because I had a high horsepower car and could turn fast laps it didn’t mean that I had moved from a good to a great DRIVER, as demonstrated by my new sports car. The high horsepower BMW M cars masked my lack of practice and expertise at my constraints. So long as I was somewhat quick around a corner as soon as the car was settled I could get on the gas and then use the car’s power to get down the straight. Then over slow the car, rinse and repeat. That was OK for the big car and I was proficient at that method. But to get much faster (like the semi-pro drivers) with the smaller, less powerful, better handling cars my lack of skills showed, and I was NOT happy.

So I hunkered down. Read books. Reviewed the basics I was taught, again. But this time I looked at it from the constraint - my slower cornering speed due to over slowing the car. I also got a coach, Spent more time in the car practicing fundamentals. Reviewed a ton of telemetry to gain feedback on where there was room for improvement, and deliberately practiced. And guess what? I got better. Much better. Way faster and far more consistent. And that low horsepower car was now turning lap times of great drivers in race cars.

Don’t expect Kubernetes (or other tools for that matter) to enable high performance

Every day, we walk into companies that either have or are about to wrongly make the decision to try and solve for their lack of experience and skill, and a desire to become a higher performing team and company by choosing to deploy a new tool. And as I mentioned above, I totally understand why: It is perceived as an easy button. This is how we’ve been sold to for a long, long time. It is programming. Can’t get where you want fast? Buy something with more power! Can’t lose that weight? Take a pill. In fact, everywhere you look it’s all about what you can do to shortcut what is really required: Getting a coach, examining your diet, and making tough choices. Transformation is a destination, not something you buy.

Specifically related to the container orchestration space (something that I am deeply passionate about) there is a troubling trend on the rise. I can count 10 Kubernetes platforms offhand, not including the three major cloud providers’ offerings. They all offer similar functionality with their own “differentiation” thrown in, whose value is questionable. Teams all over are deciding to trial, PoC, and “bake-off” platforms in an attempt to choose one that will solve their issues. And many times, these initiatives are led by the very “DevOps teams” that should be thinking differently and are being looked at to provide DevOps automation and pipelines. Well, I’m here to tell you this approach is filled with anti-patterns:

So how do you make a change to a vastly different approach before more time and money is spent without a focus on obtaining outcomes?

Takeaway: Get a coach to help you transform, not a tool

The hardest part is going to be stopping in your tracks and looking inward, rather than outward. What are the problems that you are really trying to solve? Have you quantified these? How about identifying the constraints that need to be removed and the processes which are outdated that need to be updated and optimized? These questions and the solutions that you are looking for, like developing an open source first policy, is something that a high-performance coach can help you with.

With the coach involved, you’ll soon be turning out applications with higher quality code at a higher velocity that are unshackled from the underlying infrastructure. Real improvement and transformation. Or if you are a car guy, you’ll be turning faster laps.

Insight Authors

Nebulaworks - Wide/concrete light half gray

Looking for a partner with engineering prowess? We got you.

Learn how we've helped companies like yours.