MassMutual’s Kubernetes Journey: From “Hello World” to the Real World

Lisa Martin and John Furrier were live with theCUBE at KubeCon + CloudNativeCon North America in Detroit, Michigan. During three days of coverage, they focused on Kubernetes security, large scale, and cloud native at scale.

So, what did Haseeb Budhani, the CEO of Rafay, a fast-growing private tech company and Santhosh Pasula, the global head of Cloud SRE at blue-chip practitioner MassMutual have to say?

You guys are both pioneering the next-gen cloud. Santhosh, take us through the story of your successful transformation at MassMutual.

Santhosh: The whole cloud journey in big companies – large financial institutions, healthcare industry or insurance sector – takes generations of leadership to get that level of perfection. And ideally, the strategy for cloud starts in. How do you standardize and optimize the cloud? And then, operationalization of the cloud, especially if you’re talking about Kubernetes.

In the traditional world, almost every company is running their applications in middleware. Then containerization came in. Docker is basically the de facto runtime containerization. So Docker Swarm is one of the technologies that they adopted. And, eventually, when we take it to more complicated application implementations or modernization efforts, that’s when Kubernetes plays a key role.

And as Haseeb was pointing out, we never saw so many companies working on Kubernetes. So that should tell you about how fast Kubernetes is growing and how important it is for cloud strategy.

What’s on your agenda now as you look forward?

Santhosh: We are past the stage of proof of concepts, proof of technologies, and pilot implementations. We are actually playing the real game now. In the past, I used the quote, “Hello, World to real world.” So we are actually playing in the real world, not in the “Hello, World” anymore. This is where the real-time challenges will pop up. If you’re talking about standardizing and then optimizing the cloud, you ask, how do you put your governance structure in place, how do you make sure your regulations are met, and how are you going to scale it? And while scaling, how are you going to keep up with all the governance and regulations that come with it? We are in that stage today.

“Hello, World to real world.” – Santhosh Pasula, the global head of Cloud SRE, MassMutual

Haseeb, during your customer conversations, who are the decision-makers in terms of the adoption of Kubernetes these days?

Haseeb: Well, people like Santhosh. One of the things I’m seeing here is the idea of a platform organization and enterprises. Consistently what I’m seeing is somebody—like a CTO or CIO level individual—making a decision.

I have multiple internal business units who are now modernizing applications. They’re individually investing in DevOps, which is not necessarily a good investment for their business. They think, “I’m going to centralize some of these capabilities so that we can all benefit together.” And that team is essentially a platform organization. They’re making Kubernetes a shared services platform so that everybody else can come and consume it. What that means to us is that our customer is a platform organization and their customer is a developer. So we have to make two constituencies successful. Our customer, who’s providing a multi-tenant platform, and then their customer, who’s the developer. Both have to be happy. If you don’t solve for both constituencies, you’re not going to be successful.

So you’re targeting the builder of the infrastructure and the consumer of that infrastructure on the other side?

Haseeb: Yes, it has to be both. It takes iterations to figure these things out. But this is a consistent theme that I am seeing. In fact, what I would argue is that every enterprise should be stepping back and thinking about what their platform strategy is. If you don’t have a platform strategy, you’re going to have a bunch of different teams who are doing different things, and some will be successful and some will not be. And that is not good for business.

Santhosh, you’re doing essentially what Google did at their scale, but you’re doing it from MassMutual. Is that what’s happening? And you guys are playing in there, partnering?

Santhosh: If you take the traditional transformation of the roles—in the past it was called operations, and then DevOps came in, and now SRE is the new buzzword. The future could be something like product engineering. In this journey, what worked for Google might not work for a financial company, it might not work for an insurance company. So it’s okay to use the term SRE, but at the end of the day that SRE has to be tailored down to your requirements, and the customers that you serve, and the technology that you serve.

The hard line between development and operations is thinning down. Over time, that line might disappear. And that’s where SREs are fitting in.

What’s the new thing?

Santhosh: The faster market to value that we bring to the table is very important. Think of business as an idea. How do you get that idea implemented in terms of technology and take it into real-time? We have to shorten that journey, we have to cut down time to value with technology. The technology is Kubernetes.

What used to take a year or six months can be done in a month today or even less than that. Kubernetes provides speed, velocity, agility in general, and also flexibility, as well as automation—especially if you have to maintain thousands of clusters. Today, it is possible to make that happen with a click of a button. In the past, it probably used to take a 100-person operational team to do it and a lot of time, but now automation is happening. We can get into the technology as much as possible. Blueprinting and all that, it made it possible.

How has the culture of MassMutual evolved to be able to deliver the velocity that your customers are demanding?

Santhosh: Once in a while, it’s important to put yourself in the customer’s shoes and think from their perspective.

Business does not care how you’re running your IT shop. What they care about is the stability of the product, the efficiencies of the product, how easy it is to reach out to customers, and how well you are serving the customers. So whether I’m implementing Docker in the background, Docker Swarm, or Kubernetes, business doesn’t even care about it. What they really care about is whether your development environment goes down. And if your solution is not as efficient as the business needs, then that’s a problem. So our job is to make sure, from a technology perspective, that you can implement quickly and efficiently. And at the same time, do it within the guardrails of security and kubernetes compliance.

Can we say that you’re the “vCenter of Kubernetes”, as a metaphor? A place to manage it all, one pane of glass, so to speak. Is that how you see success in your environment?

Santhosh: Virtualization has gone a long way, from what we call bare metal servers to virtualized operating systems. Now we are virtualizing applications and we are virtualizing platforms as well. And that’s where Kubernetes plays an important role.

So you see the need for a vCenter-like thing for Kubernetes?

Santhosh: There’s definitely a need in the market. Let’s say there is an insurance company who actually implemented Kubernetes and they gain the market advantage. Now, the competition wants to do that as well. So there’s definitely a virtualization of the application layer that’s very critical and it’s a critical component of cloud strategy as a whole.

“There’s definitely a need in the market…for a vCenter for Kubernetes.” – Santhosh Pasula, the global head of Cloud SRE, MassMutual

Haseeb: Maybe we should think about what vCenter does today. vCenter, in my opinion, is one of the best platforms ever built. VMware did an amazing job. They took an IT engineer and they made that person able to do storage management, networking management, VMs, multi-tenancy, access management, and audit. Everything that you need to do to run a data center, you can essentially do from a single platform.

Why are we not doing that for Kubernetes? Now you need cluster management, you need access management, you need blueprinting, you need policy management, and all of the things that have happened before, like chargebacks. So do we do many of the things that vCenter does? Yes. Are we a vCenter for Kubernetes? That is another question.

But if Rafay doesn’t move the needle, does it help productivity for developers? And does it actually scale up the enablement?

Santhosh: Yes, absolutely. Think about the hoteling industry before Airbnb and after Airbnb, or the taxi industry before Uber and after Uber. If I’m providing a Kubernetes platform for my application folks or for my application partners, they have everything ready. All they need to do is build their application and deploy it and run it. They don’t have to worry about provisioning the servers, building the middleware on top of it, and then doing testing to make sure they iron out all the compatibility issues and whatnot. Today, all I say is, “Hey, we have a platform built for you. You just build your application, and then deploy it in a development environment.” That’s where we put all the pieces of the puzzle together. Make sure the application is working, and then go to production.

And on top of it, if someone has an idea, they generally want to implement that idea very quickly. How do you do it? Don’t struggle to build an environment to implement your idea and test it in real time. From an innovation perspective, agility plays a key role—and that’s where the Kubernetes platforms or platforms like Kubernetes come in.

Andy Jassy, when he was the CEO of AWS, said that companies are going to be builders, where it’s not just utility, that you need “table stakes” to enable that new business idea. And so, in his last keynote, he did this big thing, “Think like your developers are the next entrepreneurial revenue generators.”

What do you think about a focus on developers being the product, essentially? Do you see that coming sooner than later?

Santhosh: I think it’s already happening at a certain level. Now the question comes back to reality.

You can do your proof of concept, proof of technologies, and then prove it out: “Hey, I got a new idea. This idea is great, and it’s going to give the business an advantage.” But we really want to see it in production, live, where customers are actually using it.

Haseeb: I think we’re talking about agility, but the really important thing is—these are enterprises. They have certain expectations. Guardrails are key. So it’s automation with the guardrails. Developers don’t care about guardrails. They just want to go fast. They also may bounce around a little bit off the guardrails.

One thing we know is the expectations of consumers are not going to slow down. So the ability to really make life easy for platform teams is not trivial.

Rafay and MassMutual have formed a strategic partnership to enable and empower development teams. They are on a journey together that ultimately results in turning the developer’s ideas into reality. Which means consumers see better products, and gives MassMutual a competitive advantage.