Forget the tired car metaphors — let’s try life on two wheels
I recently took a motorcycle safety course, something I have been wanting to do for a long time. The class was a great experience and taught me how little I knew about street riding. Until now, my experience has been with dirt biking offroad.
The class was “conveniently” held at a motorcycle dealership that had a small track in the back. After completing the class I fought the urge to go into the dealership, but a couple days later I stepped inside looking to buy a bike. And, uh, the rest is history…
So what does this have to do with Kubernetes? Buying a Kubernetes management platform is surprisingly similar to buying a motorcycle. Both are about getting power, control, and freedom, but each comes with choices, risks, and trade-offs.
When there are lots of options, the right dealer matters
When I walked into the dealership it was a little overwhelming, with all of the different motorcycle manufacturers and the variety of bikes. This is very similar to the current Kubernetes management landscape, where you have the choice of public cloud vendors, private datacenter players, self-managed, cloud-managed, and fully managed platforms.
Even though I did some of my own research, I benefited from the knowledge of the salesperson I worked with — just like a good value-added reseller (VAR) does in the world of software. He asked me questions about what type of bike I was looking for, how I planned to use the bike, my riding experience, and if I had a bike in mind already.
It was worth his while to give me good advice. Not only was the sales person focused on selling me a bike, he was also selling me on the dealership itself to win my ongoing business with service, parts, and accessories in the long term.
Your first decision: DIY vs off the shelf
The salesperson didn’t try to sell me parts to build a custom motorcycle myself, even though he very well could have presented that as an option.
Building a motorcycle from scratch is a bit like building your own Kubernetes clusters and self-managing them with open source tools, homegrown platforms, and processes unique to your organization.
In both cases, you learn a lot, probably have some fun with tinkering, and get complete control and customization over the end result.
The downside, of course: a lot of swearing and bloody knuckles, a lot of costly hours burned, and then you end up with all the challenges of maintaining a custom build, where even simple spares and replacement parts might be incompatible or need fabricating from scratch.
In our world, we can think of this as tech debt. If you’re building Kubernetes clusters by hand, fully customized, or choosing a Kubernetes management platform that has an opinionated implementation, you will quickly encounter technical debt issues and limitations.
Conversely, a pre-built bike from a major manufacturer might not fit you perfectly or have every feature you want, but it’s ready to ride out of the crate and is probably easy to customize over time using aftermarket parts like exhausts, mirrors and luggage, as your needs evolve (and your bank balance recovers).
Think of the Kubernetes ecosystem and the CNCF landscape as the ocean of aftermarket parts for your Kubernetes clusters. You need to be mindful of the needs and advantages of customization and variety as time goes on and technology evolves. Be careful not to choose a Kubernetes management platform that will prevent the adoption of emerging technologies due to a rigid or opinionated platform that can ultimately lead to vendor lock-in.
Look beyond the headline costs
Most dealerships use a tricky strategy where they advertise the MSRP (manufacturer suggested retail price) of the bike and hang a price tag on it. The problem is that there are a number of additional fees and running costs associated with the purchase of the bike that you may not be aware of until you dig deeper and sit down with the salesperson.
This is also the case with Kubernetes clusters. For instance, you can see the price per minute or hour advertised on a public cloud to run a managed K8s environment but you don’t see all of the other hidden costs. What kind of costs are we talking about here?
Setup costs
Dealerships charge a freight and assembly fee. Motorcycles are shipped to dealers part-built in a crate of parts, the dealership mechanics assemble the bikes and the cost of that labor is calculated as an additional fee.
When it comes to Kubernetes management platforms, you are basically presented with the same situation: you need to go through a setup and integration process, which either you need an expert team to do in-house, or you need to pay an outside vendor to do. Some platforms need more assembly than others!
Running costs and finance
Some bikes sip fuel. Others burn through rear tyres, eat chains, and guzzle gas. Some dealers hit you with a huge interest charge on finance, too. Your yearly TCO just got a lot bigger than you budgeted for.
It’s the same for K8s platforms. Some have modest usage-based pricing, discount for multi-year commitments, and ‘true forward’ out of contract usage.
Others hit you with upsell costs for ‘advanced’ features, bundle up unrelated products into enterprise agreements, or have charges for management servers and other overheads.
Work out your TCO for your expected usage before you make your decision.
Support
Motorcycles need regular maintenance, and sometimes things fail and need troubleshooting and fixing. In some cases, your warranty will cover the work, and in others you might purchase a service contract from your dealer or independent mechanic that gives you not just break/fix, but regular tune-ups, oil changes and even a wash and polish.
In Kubernetes land, you can choose to do all this yourself — troubleshooting, fixes, and regular operational maintenance — if you have the time and expertise. You can pay an MSP or outside company to do some of it for you. Or you can take advantage of the vendor’s different support offers and customer success services.
Not all support contracts are created equal, whether for bikes or software. A good mechanic or support engineer is a goldmine of wisdom. But some refuse to cover third-party components, some do poor quality work, and some will take weeks to get you a fix.
Vendor risk
Some motorcycle manufacturers, like BMW Honda and Harley-Davidson, have been around for decades and are unlikely to be going away any time soon. Others, like KTM, Husqvarna and Victory have had distinctly rocky financials, with associated concerns about product quality. Sure, you’re buying a bike today, but will the dealer be there to support you tomorrow?
Same for software vendors and even open source projects. As our 2024 State of Production Kubernetes research found, a lot of IT leaders have been burned by software they rely on being orphaned.
Insurance
As we all know, insurance is protection against the unknown and peace of mind. I want to know that if someone runs me off the road or backs into my motorcycle in a parking lot that I have some coverage to get back in the saddle.
When it comes to Kubernetes management platforms, organizations need to evaluate the protections they have to ensure business continuity. This first and foremost includes the practice of backup/restore operations for your clusters.
But business continuity includes much more, such as the mean-time-to-resolution (MTTR) which is the average time it takes a company to respond and resolve issues related to security vulnerabilities, outages (planned or unplanned), and natural disasters. As a product, a Kubernetes management platform should be architected for resilience. As a business, they should offer the level of support and SLAs (service level agreements) that align to your business continuity goals and strategies. This is all about planning for the “what if” scenarios that could impact your IT and business operations.
Choice and flexibility
The availability of aftermarket parts was important to me when choosing a bike. I wanted to make sure that I could customize my bike to my needs and have future options with parts that will be compatible with my bike — that the manufacturer will be happy to support without invalidating my warranty. I didn’t want to buy a bike that would limit me.
In the world of cloud-native, the extensibility of Kubernetes and the cloud-native ecosystem is absolutely fundamental to its value proposition. This is actually the promise of the Kubernetes project: it gives you choice. So when looking for a Kubernetes management platform you don’t want to be limited to opinionated decisions that the platform has implemented that prevent you from choosing your own modifications such as logging, monitoring, security, and developer productivity tooling. Above all, you don’t want a Kubernetes management platform vendor that will refuse to support you making the changes you need to your K8s environments.
Versatility is also important. Some bikes are built for the racetrack and are hideously uncomfortable on long journeys. Some are great off-road and suck on the highway. I chose a bike that will allow me to ride on and off-road, with or without passengers and luggage — they are known as adventure bikes.
You need to ask yourself where your Kubernetes adventure is going to take you: public cloud? Data center? Bare metal? The edge? If you’re not sure, choose a platform that can serve all of your use cases.
Protection
One thing a lot of people don’t think about when first getting into motorcycling is the cost of the riding gear. Buying helmet, jacket, gloves, boots and other gear that meets the highest standards can be incredibly costly — but could save your life.
Similarly, fitting a bike with security features like an approved immobilizer, tracker, and a good lock can help prevent theft.
It’s the same with Kubernetes. It can pay dividends to invest in a platform that’s hardened from top to bottom, with security scans and dedicated security features, regular patching, and proof of compliance to the highest security standards.
Of course, with both bikes and Kubernetes, you can have ‘all the gear, no idea’ — how you operate the technology under your fingertips really matters.
Test drives and reviews
Once I knew what type of bike I was in the market for and had done all my desk research, only then was I ready for the test ride.
It’s the same when you’re shopping for a Kubernetes management platform. Read the magazine reviews, check out user comments, then it’s time to dive in with a live demo and trial.
When you’re taking a test ride, especially of a big and powerful bike, the dealer doesn’t just throw you the keys. They’ll give you a lot of valuable info that you need to get the most from your time with the bike: specifics of the controls, idiosyncrasies with the handling, and recommendations for the best route to take to show off the performance safely.
This is even more important with a K8s management platform. Bikes have very standardized controls and architectures; that is not the case when it comes to software. Diving in unaccompanied isn’t the best way to use your time with the product — it’s why vendors make Getting Started guides and run accompanied demos with engineers on hand to help.Take advantage of these offers.
The joy of the ride
When you pick the right motorcycle for your needs and that matches your skill level, riding feels effortless — just like using a Kubernetes platform that fits your team’s workflow.
As you choose the right Kubernetes management platform for your organization, make sure the solution can grow with your business and keep pace with the changes in the CNCF landscape as well as other emerging technologies. You also need to be honest with yourself about the resources in your team and cultural factors at play in your organization.
I can’t stress enough the importance of customization, both in the world of bikes and Kubernetes. This truly is the fun part about Kubernetes, how quickly you can adopt new technologies or practices to empower your software development and delivery goals. You don’t want to limit the power of Kubernetes by choosing an overly opinionated platform that prevents innovation and the support for future operating environments that may not currently be on your radar.
Whether you are looking for a new bike or a next generation Kubernetes management platform, I hope you found this blog useful. Please be sure to check out Spectro Cloud Palette if you are in the market for a Kubernetes management platform.