Posted in

Microsoft and Anyscale Launch Managed Ray Service on Azure

by ailona•

Microsoft and Anyscale partner to deliver a managed Ray service on Azure, revolutionizing distributed AI/ML workloads. This integration simplifies scaling Python-based distributed computing with enterprise-grade Kubernetes infrastructure, enabling faster prototyping, efficient resource use, and secure, production-ready deployments.

Revolutionizing Distributed AI/ML with Azure and Anyscale

Scaling AI and machine learning workloads is often a complex, time-consuming challenge. Teams spend more hours orchestrating distributed compute than building intelligent models. Imagine if scaling AI workloads was as natural as writing Python code. This vision drives the integration of Ray, an open-source distributed computing framework, with Azure, powered by Anyscale’s managed Ray service. Together, they simplify distributed AI/ML at scale, transforming prototypes into production-ready systems with ease.
“By combining Ray’s flexible APIs with Anyscale’s managed platform and Azure’s Kubernetes infrastructure, developers can scale AI workloads faster and more reliably,” said Brendan Burns, CVP Azure OSS Cloud Native.

Unlocking Enterprise-Grade Distributed Computing on Azure

Ray’s Pythonic APIs allow developers to convert local code into distributed tasks without rewriting core logic. This means you can scale from a laptop to a large cluster seamlessly. Anyscale’s managed Ray service on Azure adds enterprise readiness through RayTurbo, a high-performance runtime optimizing cluster efficiency. With this setup, teams can quickly spin up Ray clusters without Kubernetes expertise, allocate tasks dynamically across CPUs and GPUs, and benefit from elastic scaling and fault recovery. Moreover, Azure Kubernetes Service (AKS) underpins this solution, ensuring high availability, elastic scaling, and integrated governance. AKS handles resource orchestration while providing native integration with Azure Monitor, Microsoft Entra ID, and Blob Storage. This combination guarantees secure, scalable, and compliant AI workloads inside your Azure subscription.

Practical Benefits for AI/ML Teams

This partnership removes operational barriers, letting teams focus on innovation rather than infrastructure management. You gain flexibility to prototype rapidly, optimize costs with GPU packing and spot VMs, and deploy mission-critical AI systems globally. Additionally, unified billing and compliance within Azure simplify governance for IT and data science teams alike. The result is a streamlined path from experimentation to production, accelerating AI development cycles.
“Microsoft and Anyscale are empowering developers to move faster, scale smarter, and deliver AI breakthroughs on Azure,” Burns added.
In conclusion, the integration of Ray, Anyscale, and Azure Kubernetes Service is a game-changer. It democratizes distributed AI/ML at scale with a developer-friendly, enterprise-ready platform. Tech professionals can now unlock the full potential of Python-based AI workloads, driving innovation without the usual complexity. This partnership truly empowers your AI journey from code to cloud with unmatched simplicity and power.

Key points from the article:

  • Run distributed Python workloads effortlessly using Ray’s native APIs with minimal code changes
  • Anyscale’s RayTurbo runtime optimizes cluster efficiency for accelerated AI/ML training and inference
  • Azure Kubernetes Service (AKS) ensures scalable, resilient, and secure orchestration for production workloads
  • Elastic scaling and GPU packing reduce costs while maximizing compute resource utilization
  • Unified governance and native Azure integrations streamline compliance and operational management

    • From the Source