Emerging Kubernetes tools and enterprise backup strategies for modern workloads

Bart: So, first things first: Who are you? What's your role? And where do you work?

Santosh: Hi, I'm Santosh Vallurupalli. I'm a Senior Solutions Architect with AWS. I've been with AWS for five years and primarily work with betting and gaming customers. I work deeply with Kubernetes and EKS for my customers and specialize in containers, Kubernetes, and generative AI.

Bart: Now, what are three emerging Kubernetes tools that you're keeping an eye on?

Santosh: There's a lot happening in the container space. Things are rapidly evolving, and every six months, you see a whole set of new applications emerge. Currently, I'm hooked on Kubeflow. For folks tuning in for the first time, Kubeflow is essentially a machine learning development kit. It helps you build MLOps pipelines, test deployment pipelines, inference payloads, test beds, and pretty much everything you need for machine learning. It's been my go-to place ever since I started experimenting with machine learning on EKS and containers.

A second tool I always go back to is Crossplane. If you need to manage infrastructure outside of Kubernetes—such as cloud resources, file servers, and block storages—instead of jumping onto 10 different screens and editing 100 different files, Crossplane helps you build infrastructures seamlessly and makes management really easy.

And there's no Kubernetes user without the K9s tool. It's my favorite terminal tool, which helps you access Kubernetes clusters, figure out usage patterns, navigate pods, troubleshoot, and debug.

Bart: Organizations are increasingly running mission-critical applications and apps with persistent storage on Kubernetes. What are some of the common challenges they face with respect to data protection and resiliency?

Santosh: Up until a few years ago, customers had the perception that Kubernetes was only optimized for stateless workloads. In the last five years, there has been a huge adoption of Kubernetes for stateful workloads as well. That's why many customers now run database applications, messaging applications like Redis, and enterprise messaging systems like IBM MQ and Apache Kafka in Kubernetes, as it has become highly optimized for stateful workloads.

With this adoption, customers frequently ask: How do I back up my clusters efficiently? How do I protect my persistent volumes and stateful workloads in an immutable fashion? How will I recover from a disaster recovery scenario?

Here at AWS, we have a philosophy of working backwards from customer feedback. That's why I'm excited to announce that we've launched AWS Backup Service, which is now capable of protecting EKS workloads. Starting today, customers can protect their EKS workloads, persistent volumes, dump storage backups into backup vaults, and copy backups across regions and accounts. I'm looking forward to customers trying out these new features and hearing their feedback.

Bart: Many enterprises are expanding their Kubernetes deployments across multiple regions. What strategies would you recommend for simplifying backup management at scale?

Santosh: Backup management is always a question that our customers grapple with. When customers start with Kubernetes workloads, I recommend creating backup policies for one cluster at a time. Eventually, when they reach 50 clusters, they might realize they now have 50 policies. This complexity creeps in quickly, which is why I strongly recommend a policy-driven approach to backups.

These policies would contain tagging-based strategies, dividing clusters into high, medium, and low priority. Based on these priorities, you can vary backup schedules, the number of backups, and determine whether they require cross-region resiliency. I recommend using policy-based backup policies instead of creating one-off policies for clusters individually. This approach helps you scale efficiently.

Another critical aspect is storing backups in cross-region and cross-account locations. This ensures you have a resilient posture that gets you up and running in the event of a disaster.

Bart: How are organizations balancing the need for agile container deployments with increasing regulatory requirements for data protection and retention?

Santosh: The whole philosophy of Kubernetes and its rapid explosion was driven by developers and enterprise customers wanting to ship products quickly. The faster the velocity, the higher the business value. This often put the concept of validations and maintaining application efficiency on the back burner.

Fast forward to now, with the availability of tools like OPAs, Gatekeepers, and Kyverno, customers can now maintain their application posture without sacrificing speed. They can take a shift-left strategy by integrating these tools natively into their CI/CD pipelines. At the time of deployment, they get alerted if compliance is not met or if application postures do not align with customer preferences.

I'm seeing many customers adopt this approach and I'm excited about further advancements in this space. Today at KubeCon, I'm looking forward to running demos at the AWS booth. I'm particularly excited about developments in Kubernetes, especially in machine learning and the cost optimization efforts from the open-source community.

Bart: How can people get in touch with you?

Santosh: You can reach me on LinkedIn at Santosh Vallurupalli. I'm pretty active there. For customers who want to email me, I can be contacted at svallurupalli@amazon.com. I'm quick to respond, typically within a couple of days. Feel free to get in touch—I'm excited to connect with customers.