Enterprise Kubernetes Lessons Learned from MoneyGram and HP
Many enterprises are embracing Kubernetes, but enterprise-grade Kubernetes isn’t necessarily easy. Organizations need automation and governance to make Kubernetes adoption feasible and reduce risk.
Kubernetes is proving to be a valuable part of many organizations’ digital transformation efforts, even if many of their first experiences with it are only the start of their modernization journeys.
During a panel discussion, “Kubernetes Lessons From the Trenches,” Haseeb Budhani, co-founder and CEO of Rafay Systems, Bruce Blaho, HP fellow and vice president and chief technologist of advanced solutions at HP, and Adnan Khan, enterprise architect at MoneyGram International, shared the lessons they learned from their experiences with Kubernetes adoption. Let’s take a look at some of the key takeaways and real-world experiences they shared.
Scaling with Kubernetes
During the conversation, Blaho mentioned that many companies initially focus on their solutions, not necessarily considering how production will scale as demand grows. These customers may not be aware of the challenges that come with scaling Kubernetes until the need is more immediate.
“A lot of times when we talk to these customers, they’re in the stage where [they] are running a dozen locations. As they start to scale, they think about what happens when they get to 100 stores, 1,000 stores, 10,000 stores.” This is really where the complexity trickles in. “We sometimes have to be a little patient and wait for them to realize that if they are going to run a business at this scale, they are going to need help.”
Blaho described how some customers begin the modernization process: “We see a lot of folks trying to deliver, say, computer vision-based solutions at the edge, a frictionless checkout system for a grocery store or visual analytics to count customers in a store. They’re very focused on the solution, as they should be, so the platform that they’re running on is something that they did themselves when there weren’t many commercial solutions available.”
Khan also emphasized the challenges of managing Kubernetes at scale by noting that “the operational impact of operating Kubernetes clusters at scale is not a trivial problem. You might be using a managed service like EKS, but that’s only one element of the control plane. There’s a lot more that goes into it all the way from creating clusters, maintaining them to Day-2 operations, updates, central dashboards, etc. What has worked for us is having a good partner like Rafay that is responsive to our needs and who can help us from an operational aspect but also work closely with us on the day-to-day.”
While many companies are still beginning to tap into Kubernetes and containerization to ramp up their app modernization journeys, the issue of scale is either looming or present for most. Instead of doing Kubernetes on their own, companies should look to solutions partners to help offload the burden of Day-2 Kubernetes.
Edge Computing Trends
Speaking of modernization, conversations and developments around edge computing are becoming more prevalent as businesses look to collect and analyze data in real time across hundreds or thousands of remote locations. Blaho and his team at HP are experiencing firsthand the transformation and challenges of edge, which he talked about in the discussion.
“The growth of data science and AI, and the explosion of data at the edge, has really led to an opportunity and need for edge computing,” Blaho said. “By bringing it closer to the end user and out of the data center, the core processing saves computing resources and provides a smoother, streamlined experience.”
“In this vein, many of the technologies in development for retail, health care, banking and other businesses that will use edge and on-prem Kubernetes may not be fully prepared for the realities of managing these systems. Kubernetes in the data center with a full in-house IT organization is one thing, but edge applications running at a restaurant are another matter,” Blaho said.
Businesses are finding that ideas that might have been unthinkable a few years ago are possible to quickly put into practice once they have the right supporting technologies.
Blaho and the HP team are also looking at investing more in security when it comes to edge technologies. They are particularly concerned with knowing that these highly non-secure locations are trustworthy.
He said that “being able to secure frameworks against people trying to steal data is the sort of thing we’re trying to do to let the developers focus on their magic, their innovation, and let the store owners and the operators focus on their business and get the rest of this edge stuff out of the way so they don’t have to worry about it. We are trying to build everything out with zero trust.”
Blaho mentioned that there is a big opportunity to simplify, automate and make edge computing turnkey. Edge is still relatively nascent, but it’s a technology that companies in industries like retail, finance, health care and telco will need to take advantage of to stay innovative.
Tackling Kubernetes Complexity
With many enterprises trying to do Kubernetes on their own, Budhani reiterated the importance of focusing on the value modernization ultimately promises. Companies taking on modernization should not lose sight of the potential, even as they work through the complexities of Kubernetes.
“Our customers look at the Kubernetes problem and think, ‘If only there was a magic wand to wave and the operations problem around Kubernetes goes away.’ We get to see a lot of enterprise complexity,” he said. “No question there’s pain, but it’s outweighed by the value that businesses extract.”
Added flexibility, better options and more visibility are often worth the effort and can be genuinely transformative for business models. Businesses are finding that ideas that might have been unthinkable a few years ago are possible to quickly put into practice once they have the right supporting technologies, with Kubernetes being a key enabler of better scalability once complexity is managed.
There is also complexity associated with the sheer number of tools available. Khan added, “When you go to Kubernetes, you have all of these control plane elements, logs and so many tools.
“We have a team at MoneyGram that looks at our cloud migration journey, our migration strategy… It’s a very collaborative effort. We work closely with our application architects to see what would work well within our ecosystem. We go through a round of POCs [proofs of concept]. We do extensive testing, stress testing, etc., to find what works.”
While tackling the complexity of Kubernetes can be daunting, the value reaped in the long term is worth the journey. Again, finding a partner that can help manage this complexity for you so that your developers can work on what matters most is key.
The Future of Kubernetes
Going forward, the Kubernetes ecosystem is poised to reach maturity rapidly. Budhani closed the discussion with some wisdom and predictions on where Kubernetes adoption is heading.
He noted that “application requirements necessitated by business value create a tight feedback loop between software development of new Kubernetes technologies and needs born out of market demand.” As a result, Kubernetes is marching forward quickly.
Budhani mentioned that this past year especially has reached an inflection point. “We started in this area of Kubernetes and edge computing, wondering if there’s a market there. Now it’s like, how fast can we get there? The pace at which this is going, two years from now is enough for everybody to be on the same page. If we come back and talk again in six months, I think we’ll talk about the same upsides of Kubernetes but fewer of the downsides, because the market is solving them so fast.”
Although Budhani, Khan and Blaho shared some of the challenges and complexities that inherently come with Kubernetes and their transformation journey, it’s clear that Kubernetes is worth the effort in the long run. Look for a partner to help you accelerate this journey and take away the Day-2 operational burden of Kubernetes so your developers can get back to what they love doing most — innovating new technologies and improving their applications.