Rethink Your IT Operating Mode, not your tools

One Year of “Cloud Operating Mode” conversations & observations

It’s been a bit over one year since I joined Platform9 Systems, a young and dynamic company which has a radical different approach and vision towards IT services. Don’t understand me wrong, I look back with happy memories to the achievements of my previous employers. Veeam Software changed the way we look at protecting mission-critical data and Nutanix introduced a whole new IT market with the aspiration to simplify datacenter services. I am proud that I was a part of these two movements and I am bullish about the future success of these companies. What I did notice -especially since the last 12-18 months- is that enterprises more than ever are relying on IT and have clear expectations from their IT business partners. In this new wave, these expectations are not just about a CAPEX vs OPEX discussion, nor is it about a new fancy tool or about new data optimization technologies. No, the single most important expectation is all about the interaction between consumers and IT. It’s about a behavioural change whereby the operating mode of IT is questioned in an unprecedented manner. 

It’s been a luxury to meet a lot of inspiring people within a variety of company sizes and industry verticals and my observation is that when it comes down to this new digital transformation there are hardly any differences between enterprises. Today, the in my opinion top IT delivery expectations of enterprises are:

1. Applications and data needs to be always available in a secure manner;
2. Quick time to market for new services that improve the business relation between the company and its customers;
3. Undivided attention to assist with, and drive towards the success of business objectives.

These 3 expectations can be all tied back to how IT operates. Businesses have tasted from a new IT operating mode and they have seen that frictionless, consumption based IT is possible. While I realize that it is a bit too simplistic to just hold the large public cloud providers accountable for introducing a new operating mode, I do think that the likes of Amazon, Microsoft, Google and a few others acted as the catalyst of this expectation transformation. These large IT providers introduced a set of concepts that prior to them sounded like a holy grail for organizations. Too good to be true. Now every IT professional takes it for granted that when they sign up for an AWS, Azure or GCP account they can consume the required IT services in a matter of seconds. Every IT organization will tell you that they would like to offer the same type of business benefits to their IT consumers from within their own controlled environment. Nobody questions the why of this evolution. Too often though, IT organizations and IT vendors are too focused on the what of this change. The answer to this question is not the introduction of a set of new tools, it is a behavioural change which brings me to the most important question: how. How is it possible that Amazon, Google, Microsoft and some other companies can deliver upon these expectations. You can summarize it in 3 major buckets:

1- Automation at the Heart of Everything

Public cloud vendors invest heavily in internal IP to manage the entire lifecycle of all services in a 0-touch automation fashion. Enterprises run their own IT stack with a combination of people, siloed & proprietary tools and ad-hoc scripting.

2- Cloud Platform Engineering at Scale

The biggest investment for public cloud vendors lies within their proprietary cloud control plane. This engine can only be built by hiring 100’s of highly talented engineers. Few enterprises can afford to hire this amount of talent and therefore cobble together teams and consultants to build one-off IT projects such as an enterprise cloud.

3- Operational Reliability & High Feature Velocity

Public cloud services get better every quarter. This can only happen beyond a tipping point of operational efficiency and scale. Enterprise teams cannot evolve fast enough and keep building up technical debt and therefore enterprises rely on expensive proprietary software & professional services trying to keep up with the latest capabilities.

In order for an enterprise to control its own destiny while embracing this new cloud operating mode, they’ll need to rethink the status quo followed with an in-depth analysis of the implementation details, the what.

Rethinking the status quo

A behavioural change is not something that happens overnight, it happens in a variety of stages and there definitely is not a single journey that leads to the goal of democratizing a cloud operating mode. In fact, every company will have to build out their own trajectory. The first stage though will be similar to everyone: realizing that a change in operating your IT services is a must if you want your enterprise to succeed.

Yes, some companies will be further along on their transformation journey, but I’ve noticed that the below questions often feed a good, business centric conversation especially if they just started on this journey:

1. Why did you select Microsoft 365 or Google G Suite instead of deploying collaboration tools in your datacenter(s)?
2. Why do you spend a lot of your IT budget towards people that -most likely- just keep the lights on?
3. Is there any business value to spend thousands of dollars and many hours of your best people’s time to upgrade IT management tools?
4. How long does it take your organization to fully ramp up a newly hired developer?
5. Do you want your developers to build high quality software services and let them focus on the code lifecycle, or are there any business reasons why you want them to also operate the lifecycle of multiple tools?
6. Do you agree that large IT providers (eg. Microsoft, Google and Amazon) have changed the way IT can be consumed and if so, do you believe that your technology incumbents such as VMware, RedHat, NetApp,... can help you to offer the same consumption experience from within your own controlled environments?

Implementing a Cloud Operating Mode. It’s not about tools!

Too often conversations around implementing a cloud operating mode are centered around the tools, software packages and professional services an enterprise needs to purchase. Ask yourself this question: How many proprietary tools did you buy in the last 12-18 months with the promise of having a stack in your datacenter that offers a real cloud operating mode/system? Hopefully these tools delivered upon the promise, but I’m afraid that the majority of promises are still a fata morgana.

I will be the last person to say that you do not need to buy services and/or tools to change your operating mode but I would recommend to always trade-off the selection of your partners against these 6 core themes:

1. Will I be able to leverage any infrastructure type?
2. Will I be able to embrace any infrastructure provider?
3. Will I be able to deploy any application runtime? (bare metal, virtual machines, containerized and even serverless services)
4. Will I be able to reduce my operational IT tasks to almost 0?
5. Will I be able to avoid lock in?
6. Will I be able to take advantage of the innovation happening in a variety of open source communities? (eg. CNCF, Apache, OSF, LF,...)

Another situation that often comes up when customers are “too deep in the tools rathole” is when customers believe that Kubernetes will solve all their delivery and operating challenges. I am always very clear on this: Kubernetes is not the solution to deliver a cloud operating mode. 

For the folks who do not know what Kubernetes is, Kubernetes is one of the most popular open source orchestration frameworks for automating application deployment, scaling, and management. It was originally designed by Google, and is now maintained by the Cloud Native Computing Foundation. It aims to provide a "platform for automating deployment, scaling, and operations of applications across clusters of hosts".

There are at least 50 “tool vendors” who are doing something with Kubernetes. Some vendors such as RedHat and VMware have even put Kubernetes at the heart of their products. You can group most of these solutions into 3 big Kubernetes deployment buckets:

1. Do-it-Yourself. The source code of Kubernetes is available online, so you can go ahead and do everything on your own;
2. Every major cloud provider has a Kubernetes service that will work beautifully on their own infrastructure;
3. You can select one of the commercial Kubernetes offerings. Some people will call it a Kubernetes fork, a wrapper around Kubernetes or vanilla Kubernetes with professional services. What is in a name; it is a software deployment + support + professional services.

While Kubernetes at some point in your journey will become important, Kubernetes on its own cannot be the cornerstone of your cloud operating mode and here is why. 

I have a positive mindset and I want to believe that the above three deployment options will provide you a decent amount of benefits such as the possibility of introducing a DevOps, GitOps, SheetOps,... culture or the possibility to offer your developers a rock solid application orchestrator. These benefits though are primarily the result of the Kubernetes community and not the deployment model. For all of these deployment options you will need to attract Kubernetes experts or spend lots of dollars to re-educate your operation teams to manage a highly complex tool such as Kubernetes. These three deployment options will eventually lead to a Kubernetes Paradox. One of the biggest reasons why organizations are interested in Kubernetes is because of how much automation they are able to drive through K8s and ultimately do more with less, but if you try to operate and manage K8s in production with people and traditional tooling you will end up with investing in more people to keep the services up and running.

And while Kubernetes is one of the hottest items in the datacenter space these days, I do think that in a few years from now the open source community will re-innovate itself again and who knows what type of tools will be made available. Therefore I would like to introduce to you a fourth deployment that I believe is the key to success when deploying a cloud operating mode.

Cornerstone of an Open Cloud Operating Mode

Independent of the technology partner you want to work with to implement a cloud operating mode, I strongly believe that there is only one possible delivery vehicle to successfully transform your IT operations. I was happy to see that Gartner recently introduced an “as-a-service implementation model” quadrant which contains a new category: SaaS Managed Clouds. 

“SaaS Managed” might sound like just another buzzword, but if you break down the concept you will quickly realize that a managed SaaS-based control plane is the key to unlock the holy grail of a cloud operating mode that respects the 6 core themes described earlier in this post.

So what is a managed SaaS-based control plane? In the most simplistic way, it is a 3rd party cloud based engine that integrates with any of your infrastructure resources (data centers, edges, public clouds, MSPs, virtualized, non-virtualized, containerized,...) to let you consume the required services to serve your IT consumers vs having to manage the life cycle of the services on your own - or with professional services. The two key elements are:

1- Any of your infrastructure resources

Only a few enterprises will want to embrace a new operating model without the possibility to onboard already existing/purchased resources. In my view, an open SaaS-based control plane should offer you the possibility to onboard any infrastructure without any strong opinions on specific hardware technologies.

2- Services that power your business applications

Equally important to the freedom to onboard any infrastructure resource, you should also be mindful that the control plane has a plug-n-play model based upon your needs. For example, some customers will want to deploy a self-service portal on top of virtualization, while other companies might want to build a bare-metal cloud. Also from a higher level point of view you need to ensure that there is zero lock-in into the offered services nor into the control plane itself. The offered services should ideally be backed by an enterprise grade SLA so you can be 100% sure that you can focus on your business services and do not have to worry about what powers your IT services.

Closing

Only a few companies have been able to deliver a managed SaaS-based control plane. Amazon, Google, Microsoft built such an engine that works well within their own controlled infrastructure stack. Most recently VMware, a well respected company, also released a SaaS-based control plane for one of their proprietary software services.  

All of this taken into account is the reason why I am a strong believer in the success of a company such as Platform9 Systems. This team figured out how to build a SaaS-based control plane that fully aligns with the 6 described themes.

I would like to thank my colleague and friend Andy Gysbrechts to push me to write down these thoughts and to review this piece.

Feel free to share your thoughts and ideas on what you believe is important for enterprises that want to start adopting a cloud operating model.