DanaCozmei summarizes Ignite 2025’s key Azure Compute breakthroughs, focusing on practical infrastructure features for scaling AI, analytics, and distributed workloads with resiliency, automation, and performance.

Scaling Azure Compute for Performance: Innovations from Ignite 2025

Ignite 2025 revealed the next wave of Azure Compute advancements to meet the real-world pressures of modern workloads—especially AI inference, data-centric analytics, and globally distributed services. This summary by DanaCozmei highlights practical improvements released and in preview:

Direct Virtualization – Immediate Device Access for High-Performance Apps

  • Enables nearly bare metal access to NVMe disks and GPUs.
  • Ideal for latency-sensitive workloads such as AI inference and gaming.
  • Isolation for child VMs hosting hostile workloads.
  • Lowers latency and cost for demanding applications.
  • Available in limited preview: Sign up for Direct Virtualization Preview

Large Container Sizes – Accelerating Compute-Intensive Applications

  • High vCPU and memory configurations to supercharge AI/ML training and analytics.
  • Allows rapid scaling of inference workloads.
  • Simplifies container orchestration: fewer containers per workload.
  • Reduces inter-container latency.
  • Generally Available: Learn more about Large Containers

VM Applications – Streamlining Global Application Deployments

  • Manage and deploy up to 25 applications per VM (2GB each).
  • Consistent deployments across regions—automatic replication and updates.
  • Automate updates without manual intervention.
  • Reduces operational overhead for global distributed workloads.
  • VM Applications GA Announcement

Scheduled Actions – Automating Operations at Scale

  • Schedule power operations for up to 5,000 VMs simultaneously.
  • Built-in reliability and safeguards for large-scale automation.
  • Actions available: Start, Stop, Hibernate (more coming soon).

Azure Compute Gallery – New Features for Reliability

  • Soft Delete (preview): Protect VM images from accidental deletions.
  • Zonal Redundant Storage (ZRS): Default for image version storage to improve uptime and resiliency.
  • Enhances operational reliability and compliance for teams managing VM images and artifacts.
  • Resiliency Announcement

VMSS Instance Mix – Flexible Capacity Acquisition

  • Define up to five VM SKUs in a deployment for capacity fungibility.
  • Supports allocation strategies (CapacityOptimized, LowestPrice, Prioritized).
  • Enables agility and cost optimization for unpredictable workloads.

Best Practices Shared at Ignite

  • Use latest SKUs for optimal performance and cost-efficiency.
  • Instance mix helps acquire capacity at scale across VM sizes.
  • VM Apps improve reliable global app delivery.
  • Scheduled actions streamline power state management (great for Virtual Desktop scenarios).
  • Build resiliency and security natively into compute architecture.

Session On-Demand

Check out the full session: BRK173: Scaling Azure Compute at Ignite

Conclusion: Enabling AI-Ready, Resilient Cloud Infrastructure

Azure Compute’s evolution reflects customer demand for intelligent, adaptive, and automated infrastructure. With innovations like Direct Virtualization, large containers, VM Applications, and capacity-optimized scale sets, enterprises can scale confidently for AI, analytics, and distributed workloads. These advancements empower technical teams to deliver robust, production-ready solutions that keep pace with the industry’s growth toward intelligent, distributed systems.

Learn more at Azure Compute Blog.

Author: DanaCozmei (Joined November 14, 2023)

This post appeared first on “Microsoft Tech Community”. Read the entire article here