Cloud Native Weekly: Istio 1.25.0 Officially Released

Open Source project recommendations

Dstack

Dstack is an open-source AI compute management platform designed to simplify the deployment and management of AI tasks. It supports running AI workloads both locally and in the cloud and provides automated GPU resource scheduling, enabling developers to utilize compute resources more efficiently. Dstack is Kubernetes-compatible and can be seamlessly integrated into existing infrastructure, making it suitable for AI model training, inference, and MLOps workflows.

SkyPilot

SkyPilot is an open-source cloud task scheduling and optimization platform that helps users efficiently run AI training and compute tasks across multi-cloud environments. It supports automatic selection of the optimal cloud provider, intelligent compute resource allocation, and offers cost-effective, high-performance compute optimization solutions. SkyPilot is ideal for AI research, distributed computing, and large-scale cloud task management, and integrates seamlessly with Kubernetes and various cloud platforms.

Kaito

Kaito is an AI-powered search engine designed to help users retrieve and organize information more efficiently. It combines large language model technology with search capabilities, allowing users to extract key information from multiple data sources (such as GitHub, academic papers, websites, etc.), enhancing research and development productivity. Kaito is suited for developers, researchers, and knowledge workers, offering a smarter search experience.

RagApp

RagApp is an open-source framework for building RAG (Retrieval-Augmented Generation) applications. It helps users create intelligent Q&A and knowledge retrieval systems based on large language models. RagApp supports extracting information from multiple data sources (such as documents, databases, APIs) and integrates with LLMs to provide accurate, context-based responses. It is suitable for enterprise knowledge management, chatbots, and intelligent search applications.

Technical recommendations

Best Practices for Efficiently Managing AI/ML Workloads in Kubernetes

This article outlines best practices for running AI/ML workloads efficiently on Kubernetes. It emphasizes the importance of managing compute resources effectively by using device plugins (e.g., NVIDIA plugin) for optimized GPU allocation and configuring CPU/memory requests to avoid contention. Techniques like node selection, affinity rules, and autoscaling can improve scheduling flexibility and efficiency.

To ensure security in multi-tenant environments, the article recommends using namespaces and network policies for isolation and enforcing access control via RBAC to prevent unauthorized access.

Monitoring and log management are also critical for system stability. Integrating Prometheus enables real-time resource monitoring to optimize performance and identify bottlenecks early, while centralized logging enhances troubleshooting efficiency. By following these practices, organizations can better manage AI training and inference workloads on Kubernetes, boosting resource utilization and system reliability.

K8s 1.31: Cloud Controller Manager Initialization Challenges & Solutions

This article discusses major architectural changes in Kubernetes 1.31 related to cloud provider integration. The update removes in-tree cloud provider code and shifts to using the Cloud Controller Manager (CCM) to handle cloud-specific logic. While this improves Kubernetes’ scalability and compatibility with cloud platforms, it introduces a “chicken-and-egg” problem: when the kubelet starts and registers the node with the API server, critical cloud metadata (e.g., address, zone labels) is missing — yet this data must be filled by the CCM, which depends on the node being initialized.

To resolve this, cluster admins and installation tools (such as kOps or Cluster API) need to implement extra steps to ensure the CCM is correctly configured and coordinated with other components during cluster initialization. By optimizing startup processes and scheduling strategies, the impact of this migration can be reduced, enhancing cluster stability and manageability.

What’s new in cloud native

Istio 1.25.0 Official Release: Enhanced Ambient Mode & Traffic Management

Istio version 1.25 introduces several key updates and enhancements, boosting observability, security, and scalability. The new version improves traffic management and optimizes sidecar resource usage. It also enhances support for the Kubernetes Gateway API.

Furthermore, Istio 1.25 strengthens authentication and access control mechanisms, adding more policy configuration options to increase overall security. It also improves observability by enhancing logging and metrics collection, helping operators monitor and manage the service mesh more effectively.

Dapr v1.15 Official Release

Dapr version 1.15.0 has officially been released. This version introduces several new features and improvements, including the promotion of the Scheduler service to stable status. The Scheduler service now manages Actor reminders by default, replacing the previous Placement service.

Upon upgrading to Dapr 1.15, existing Actor reminders will automatically migrate from the Placement service to the Scheduler service. Additionally, the release includes other enhancements and fixes aimed at improving the development experience and performance of distributed applications.

About KubeSphere

KubeSphere is an open source container platform built on top Kubernetes with applications at its core. It provides full-stack IT automated operation and streamlined DevOps workflows.

KubeSphere has been adopted by thousands of enterprises across the globe, such as Aqara, Sina, Benlai, China Taiping, Huaxia Bank, Sinopharm, WeBank, Geko Cloud, VNG Corporation and Radore. KubeSphere offers wizard interfaces and various enterprise-grade features for operation and maintenance, including Kubernetes resource management, DevOps (CI/CD), application lifecycle management, service mesh, multi-tenant management, monitoring, logging, alerting, notification, storage and network management, and GPU support. With KubeSphere, enterprises are able to quickly establish a strong and feature-rich container platform.

To stay updated, visit our official website or follow us on Twitter.