CNCF-hosted Co-located Events North America 2024

Despite DevOps being the norm and best practice for software development today, many organizations still treat their data differently from code, neglecting these well-established best practices over many years. This session will challenge this outdated approach and advocate for defining better “Data as Code” strategies, emphasizing the importance of applying the same rigor and methodologies used in software development to data operations. This session will teach how to run database services as part of Kubernetes infrastructure, just like current microservices, and apply the same SDLC methodologies to data operations as we do to code management with a common cloud native stack. In this session attendees will learn how to bridge the gap with a set of powerful CNCF Tools, including managing databases in K8s, how to embrace data as code, good GitOps practices and more.

Stream processing and data analytics are crucial for various roles, including data engineers, platform engineers, DevOps, and many more. Simplifying real-time stream data processing while ensuring it's cost-efficient and resilient to K8s pod restarts or node upgrades remains a significant challenge. While existing stream processing solutions exist, they are often hard to manage, operationally demanding, and expensive. This presentation will share our journey in developing a generic open-source Kubernetes-native stream processing framework, it allows developers to effortlessly and quickly execute large-scale stream processing tasks without the burden of heavy and costly data processing platforms.

As businesses increasingly operate in multi-cloud and multi-region environments, managing databases across these distributed landscapes presents unique challenges and opportunities. This presentation will provide a comprehensive guide to building a multi-cloud, multi-region Database-as-a-Service (DBaaS) on Kubernetes. We will discuss strategies for deploying and managing distributed databases across multi-cluster setups that ensure high availability, data redundancy, and seamless disaster recovery. Key topics include setting up multi-cloud Kubernetes clusters, leveraging cloud-native database solutions, and automating database operations using Kubernetes Operators.

In this panel discussion with leaders in banking, insurance, and technology, we will explore the convergence of Kubernetes, GPUs and AI in data operations focusing on emerging trends in the financial services industry.

Topics:

• Emerging trends in cloud-native data operations for financial services 
• How Kubernetes enables use of AI workloads 
• Future innovations at the intersection of Kubernetes, GPUs, and AI
• Challenges and solutions in adopting such technologies in highly regulated environments and what is needed to overcome them

As GPUs become increasingly powerful, the separation between compute and storage often results in underutilized GPUs waiting for data. Meanwhile, high-performance components on GPU machines, such as NVMe storage and fast networks leveraging InfiniBand or special NICs, remain idle. Effectively leveraging these hardware resources to address GPU underutilization is a critical challenge. In this talk, we introduce a Kubernetes-native distributed caching layer that leverages NVMe disks and fast networks to optimize PyTorch training data access. Utilizing stateless workers for scalability and ETCD for membership services, this caching layer efficiently manages and serves data. Cached data is rapidly and efficiently fed into GPU memory using NVIDIA's DALI data loader, GPUDirect Storage (GDS), and Remote Direct Memory Access (RDMA), significantly reducing data transfer bottlenecks and improving overall training performance.

Running PostgreSQL on Kubernetes is becoming increasingly popular, offering various methods for implementation. This talk will examine different operators that facilitate running PostgreSQL on Kubernetes. We'll also take a look at Neon PostgreSQL, specifically designed for cloud-native environments, and cover the most important best practices for effectively running PostgreSQL on Kubernetes.

Are you a systems admin or “infrastructure person” who suddenly had “looking after databases” added to your remit? This seems to happen a lot when everything’s running on Kubernetes and the databases are seen as just another part of the landscape. If so, what do you do? Quickly learn to be a database administrator? Hope the databases look after themselves? Or maybe implement a Kubernetes Operator, that has been created by database experts, to act as your virtual DBA? There are lots of moving parts to an enterprise scale database deployment, whether it’s on Kubernetes or not. You need high availability, disaster recovery, monitoring, upgrades… An Operator can do all of this for you, automating the initial deployment of a highly available database architecture, day to day management and monitoring of your databases, and even upgrades. We’ll look at what you want from an Operator, what you need to think about when choosing an Operator, and how easy it can be to get up and running.