AI-powered resiliency for modern workloads with Mark Russinovich | BRK452
Mark Russinovich, Enrique González Bejarano, Amit Ganguli, and Surya Sripathi Raju discuss practical ways to engineer Azure applications for failure as organizations modernize and adopt AI. The session combines lessons from Azure outages with resiliency planning, AI-assisted assessment, and recovery capabilities.
Engineering for failure
The session opens with lessons from Azure outages, including the 2014 Azure Storage outage and the creation of a safe deployment policy. These lessons frame the discussion of standardized deployment practices, customer trust, and why resiliency remains important as organizations adopt AI.
The speakers cover the shared responsibility model and planning for regional failures. Key areas include:
- Identifying critical application and infrastructure dependencies.
- Assessessing resilience gaps.
- Defining service level objectives (SLOs).
- Using AI-driven monitoring to support resilience.
- Building multi-region applications and strengthening business continuity.
AI-assisted resiliency assessment and automation
The session introduces Azure resiliency tooling, a resiliency platform, AI agents, and unified resiliency experiences. It also explores the role of AI agents in resiliency engineering.
A demonstration focuses on an AI-powered resiliency agent for zonal resiliency assessment and infrastructure automation, connecting resilience assessment with infrastructure modernization.
Backup and recovery
The final segment covers backup, cyber recovery, a customer case study with Santander, and new Azure resiliency capabilities. These topics complement the architecture and deployment discussion by addressing recovery and preparing applications and data to withstand disruption.
Session chapters
- 00:00 — Introduction to AI-powered resiliency and Azure outage lessons
- 00:47 — 2014 Azure Storage outage and creation of safe deployment policy
- 03:13 — Why resiliency and customer trust matter more in the AI era
- 04:20 — Shared responsibility model and planning for regional failures
- 06:23 — Defining service level objectives and AI-driven monitoring for resilience
- 08:27 — Standardized deployment practices and Azure resiliency tooling
- 11:08 — Building multi-region resilient applications and using AI agents for resiliency engineering
- 15:28 — Azure resiliency platform, AI agents, and unified resiliency experiences
- 18:10 — Demonstration of AI-powered resiliency agent for zonal resiliency and infrastructure automation
- 28:32 — Backup, cyber recovery, Santander customer case study, and new Azure resiliency capabilities
Resource
This Microsoft Ignite 2026 session is BRK452, an advanced-level (300) Cloud & AI breakout presented in English (US).