Cloud Native Weekly:Optimizing Continuous Deployment at Uber
Open Source project recommendations
Cog
Cog is a tool that packages machine learning models into containers. By configuring it, you can automatically package the environments and dependencies required by the machine learning model into a container for easy deployment, saving you the trouble of writing Docker files and dealing with CUDA. It also automatically starts an HTTP interface service for convenient invocation。
KnowStreaming
KnowStreaming is a powerful platform for monitoring and operations management of Kafka clusters。
Opserver
Opserver is a lightweight monitoring system developed using .Net. It can monitor various services including servers, logs, SQL Server clusters, and Redis, and supports customizing dashboard displays by modifying JSON configuration files。
Litellm
Litellm is a tool that simplifies the invocation of large model APIs. This project can unify the interfaces of various AI large models and services into the OpenAI format, simplifying the work of switching and managing between different AI services/large models。
Technical recommendations
Step-by-Step Guide to Deploying ML Models with Docker
This article introduces how to deploy a machine learning model using Docker, providing detailed step-by-step guidance. The process includes setting up the environment, building the model, creating dependency files, writing a Dockerfile, building a Docker image, running the container, and pushing the container to Docker Hub. Using Docker ensures consistency in model operation across different environments and stability in dependency relationships, making the deployment process smoother and more scalable, while eliminating the need for complex environment setups。
Optimizing Continuous Deployment at Uber: Automating Microservices in Large Monorepos
Uber shared in its recent article how it optimized the continuous deployment of microservices within a large-scale monorepo. Facing a microservice architecture consisting of thousands of services, Uber redesigned its continuous deployment system, “Up CD,” to enhance automation and security. This system is tightly integrated with its internal cloud platform and observability tools, achieving standardization and repeatability in deployments by introducing a unified commit process and security checks at each stage。
Since its launch, the proportion of automatically deployed services has increased from less than 10% to nearly 70%, and the rate of production incidents caused by code changes per thousand has been reduced by more than 50%. The success of Up CD lies in its user-friendly interface and strategy of phased deployment for shared code changes, effectively addressing the challenges in a monorepo and significantly improving the safety, efficiency, and transparency of deployments。
A tale of two transformations: Lloyds and Chase UK on driving change in the banking sector
Two UK banks with vastly different historical backgrounds — the long-established Lloyds and the cloud-native newcomer Chase UK — shared their tools, technologies, and philosophies driving digital transformation. Lloyds, as an institution with a 259-year history, is shifting towards a cloud-first strategy while facing the challenge of a complex technical legacy. It has achieved self-service and reuse by building an Internal Developer Portal (IDP)。
On the other hand, Chase UK, which has been cloud-native since its establishment in 2021, rapidly accumulated two million customers. By transitioning from a tool-centric approach to an SDLC (Software Development Life Cycle) ecosystem, it successfully migrated over 500 microservices within six weeks, demonstrating the future trend from tools to ecosystems。
What’s new in cloud native
Argo Workflows 3.6 Release Candidate Version
Argo Workflows version 3.6, now available as a release candidate, introduces 53 new features, 237 fixes, 164 documentation updates, and has engaged 130 contributors, with 83 being new contributors. Key updates include multiple schedules, stopping policies, and conditional expressions for Cron Workflows, various UI improvements such as displaying full workflow names, toggling time displays, Markdown support, and enhancements in metrics, controllers, plugins, storage, and performance optimizations. Users can update workflow templates like Cron Workflows through the CLI and enjoy additional convenient features.
Kmesh v0.5 Released! Advancing the Sidecarless Service Mesh
The Kmesh project team announces the official release of Kmesh version 0.5.0!
In version 0.5.0, we have implemented several significant enhancements, specifically including:
- The introduction of the powerful command-line tool kmeshctl, providing users with a more convenient operational experience.
- The achievement of more comprehensive end-to-end test coverage, ensuring the stability and reliability of system functions.
- Significant improvements in the visualization of underlying eBPF information, enhancing users’ monitoring and debugging efficiency.
- Enhanced observability, helping users more easily grasp the system’s operational status.
Provided complete restart support, ensuring the — system can quickly resume normal operation after unexpected interruptions.
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.
