Thursday marked the final full day of the Ignite conference. While a handful of sessions remained on Friday morning, the event was clearly winding down. It was also the last day the Hub was open, with sponsors and partners available for conversations and networking—making Thursday the final opportunity to absorb as much knowledge as possible.
One of the perks of attending Ignite is that a certification exam is included with the conference pass. I decided to take advantage of this “free” exam and scheduled mine for first thing Thursday morning. Having spent much of the past year attending official Microsoft Azure training classes, I felt prepared to tackle the AZ-104: Azure Administrator Associate exam. In the evenings leading up to the test, I studied and reviewed material at the hotel.
Although the exam was scheduled for four hours, it rarely takes that long in practice. Once seated and signed in, I began the exam—only to find the questions more challenging than expected. My confidence dipped early on, but I slowed down, carefully read each question, and answered to the best of my ability. After about an hour and a half, I completed the exam and submitted my answers.
The result: PASS.
Given how difficult the questions felt, the outcome was a pleasant surprise and a great way to start the day. With that behind me, it was time to head to my first session.
The New Industrial Frontier
The morning session, The New Industrial Frontier, focused on how AI is reshaping industrial environments. Early integrations of devices with large language models (LLMs) were primarily used for data collection and telemetry—feeding dashboards, generating reports, and triggering alerts when issues were detected. This marked the initial phase of sensors replacing manual processes.
We are now entering the next wave of industrial AI. Today’s systems can interpret events and respond in real time. Tools are integrated with context-aware agents capable of predicting and preventing failures, automating parts management and logistics, and even scheduling repairs. These agents reason through problems, validate decisions, and perform supervisory actions such as making recommendations or approving changes.
Microsoft has introduced specialized Azure tools designed to support industrial AI workloads, including:
- Azure Event Grid – Messaging and event handling
- Azure IoT Hub – Device management
- Azure IoT Operations – Edge operations
Together, these services work as a unified platform, positioning Azure as a single control plane for industrial environments.
ABB followed with a presentation on how they enable industrial customers to adopt AI effectively. They see AI adoption across employees, business processes, and products and services. Their solution, GenIX, offers thousands of prebuilt analytics and generative AI capabilities. When deployed in industrial settings, GenIX can identify root causes of failures and provide actionable recommendations.
Wadlebots NOVA also presented on the evolution of robotics. Many industrial robots still in use today date back to the 1980s and rely on PLCs with little to no AI or contextual awareness. Wadlebots NOVA is developing next-generation robotics integrated with NVIDIA Omniverse, simplifying configuration and bridging virtual and physical environments. Combined with Azure and NVIDIA, this approach delivers highly scalable and flexible infrastructure. The pace of innovation in industrial AI and robotics is clearly accelerating.
Architecting for Resiliency on Azure
As impressive as these advancements are, they mean little without resilient infrastructure. That made Architecting for Resiliency on Azure Infrastructure a timely afternoon session—especially coming just weeks after a major Microsoft 365 outage.
Although Azure includes built-in resiliency, the session highlighted several common customer challenges:
- Shared responsibility confusion – Uncertainty around where Microsoft’s responsibility ends and the customer’s begins
- Cost prioritization – Resiliency sometimes sacrificed for savings
- Lack of risk awareness – Misconfigurations or incomplete setups
- No measurable metrics
- Inconsistent adoption of best practices
Even with layered protections, small oversights can create “leakages” that lead to outages, as seen in the recent Microsoft Front Door incident. Ultimately, resiliency is a partnership between Microsoft and the customer.
Azure resiliency begins at the geographic level, where each geography contains one or more regions. Each region consists of three physically separate availability zones with independent power, cooling, and infrastructure.
Connectivity resiliency depends on customer choices:
- Standard (ExpressRoute) – Single circuit, single connection
- High (Metro) – Single circuit, dual location, 99.9% availability
- Maximum (Microsoft-recommended) – Dual ExpressRoute circuits, highest resiliency, roughly double the cost
To help customers evaluate their setup, Microsoft announced ExpressRoute Resiliency Insights, now GA as of Ignite. This tool identifies resiliency gaps and recommends corrective actions.
Additional resiliency features that became GA at Ignite include:
- Zone-Redundant NAT Gateways
- Now the default configuration
- Up to 100 Gbps throughput
- 5 million packets per second
- Standard Load Balancer
- Redundant by default
- Supports multiple availability zones
Azure DNS services were also highlighted as fully redundant by default, requiring no additional configuration from customers.
For virtual machines, resiliency options vary based on workload type:
Stateful workloads
- Azure Backup (low cost, slower recovery)
- Azure Disk and Zone DR
- Regional Azure Site Recovery (ASR)
Stateless workloads
- Active/Passive zones
- Active/Active zones
- Regional Active/Active with self-healing
Microsoft also introduced the Business Continuity Center, which provides application-level reporting, validated drills, and testing. This Resiliency as a Platform (RaaP) offering includes guided experiences powered by Copilot and is designed to simplify best-practice adoption.
Overall, Microsoft appears to have listened to customer feedback around complexity and lack of guidance. As noted during the session, few administrators have the time to sift through hundreds of documents and KB articles. Centralized metrics and guided configuration should significantly improve real-world resiliency outcomes.
Cloud-Native Innovations with Mark Russinovich
The final session of the day—Cloud Native Innovations with Mark Russinovich—was highly recommended by an industry colleague. Mark focused on container advancements and reinforced the idea that the future of cloud computing is increasingly serverless.
Microsoft’s Azure Container Instances (ACI) have evolved significantly. Previously, containers ran in isolated Hyper-V environments on physical hosts. They now run natively on physical Hyper-V servers, greatly improving performance. Microsoft has also introduced “standby instances,” enabling faster scaling during sudden demand spikes.
Several upcoming and recently introduced features include:
- Stretchable ACI Containers – Define minimum and maximum resource limits, allowing containers to scale up or down dynamically without spawning new instances
- Azure Managed Cilium – Replaces BYO Cilium, improving performance and observability without customer-managed infrastructure
- eBPF integration – Eliminates IP tables, consolidates policies, and boosts performance
- Azure Container Storage – Platform-managed storage with NVMe caching
- A demo showed download times dropping from over two minutes to under 30 seconds
Security improvements were also highlighted. Containers often run with minimal restrictions, but OSGuard now enables locked-down containers. In a demo, unapproved applications and commands were blocked—even via CLI—preventing misuse if a container is compromised.
Patching containers has long been a pain point, as vulnerabilities typically require rebuilding and redeploying entire images. Project Copacetic addresses this by enabling hotfixing, reducing patch timelines from days or weeks to just hours.
Mark also discussed Azure Incubations, the group he oversees, where experimental and open-source projects are developed into enterprise-grade solutions. Notable projects include:
- Radius
- Separates applications from infrastructure
- Enables portable containers across Azure, AWS, and GCP
- Uses “recipes” to bind apps to environments
- Enforces governance while preventing container jailbreaking
- Drasi
- Monitors container state
- Triggers actions based on state conditions over time
- Operates on a model of Source → Continuous Queries → Reaction
Final Thoughts
This session effectively wrapped up my time at Ignite. I had planned to attend one final breakout, but traveling from Moscone West to South—combined with security checkpoints—meant the session was already halfway over upon arrival.
As always, Ignite delivered an incredible amount of valuable information. While the conference leaned heavily into AI—perhaps more than some would like—it reflects the current direction of the industry. Overall, it was a worthwhile and insightful experience, and I’m already looking forward to what next year brings. If you are interested in an overall review of the conference, then check out the articles below.
Leave a Reply