Azure news and engineering summaries

A former Azure Core engineer details the internal mismanagement at Microsoft, focusing on the unrealistic plan to port complex Windows components to limited SoC hardware. The narrative highlights systemic bloat, poor architectural decision-making, and organizational disconnects, which the author argues have jeopardized trust in Azure’s mission-critical infrastructure and impacted key high-profile customer relationships.

GitHub’s August 17 outage lasted 7 hours and 47 minutes, disrupting authentication, GitHub Actions, APIs, pull requests, issues, and Copilot. A Central US infrastructure component failed to scale during peak traffic, causing cascading capacity pressure and retry storms. GitHub is adding hardware, expanding Azure migration, improving observability and rollouts, isolating dependencies, and standardizing retry controls.

The 2026 Cloud VM benchmarks evaluate 44 instance types across seven providers. AMD EPYC Turin emerges as the performance leader, while Intel Granite Rapids shows improved stability. For value, Oracle and Hetzner dominate on-demand charts, whereas Azure Cobalt 100 and Google Axion offer competitive ARM performance. Recommendations highlight upgrading to modern CPU generations to maximize efficiency and cost-savings.

Researcher 'Nyxgeek' discovered four Azure Entra ID sign-in log bypasses since 2023. These vulnerabilities allowed attackers to validate credentials or obtain active tokens without triggering security logs, potentially hiding malicious activity. The issues, caused by simple parameter manipulation, were reported to Microsoft and subsequently patched. Nyxgeek highlights concerns regarding Microsoft's inconsistent security bounty practices and notification standards.

PlanetScale is now available on Microsoft Azure, initially supporting PlanetScale Postgres and Neki databases in centralus, eastus2, and westeurope; Vitess is not yet supported. Azure deployments include high availability across three availability zones, Query Insights, schema recommendations, automated backups, and Metal clusters. PlanetScale built custom Kubernetes networking that assigns pod IPs in /28 blocks, increasing capacity 16x to over one million pod IPs per network. Existing databases can be migrated through support.

Broadcom reportedly removed public downloads of VMware’s VDDK, a library used by Azure Migrate, Red Hat Migration Toolkit, Nutanix Move, and other VMware migration and protection tools. Support reportedly says access is restricted, creating obstacles for customers leaving VMware. Agent-based, storage-assisted, and Proxmox’s built-in import options remain alternatives while vendors investigate workarounds.

Floci offers fast, MIT-licensed local emulators for AWS, Azure, GCP, and OCI, requiring no cloud accounts or credentials. Its native binaries support 119 AWS services and other cloud APIs, enabling realistic local development, AI-agent testing, CI pipelines, and infrastructure-as-code dry runs through a unified CLI and dashboard.

Microsoft has released Azure Linux 4.0 in public preview, transitioning it from a specialized container host to a general-purpose Linux distribution. Now based on Fedora with declarative overlays, it features improved tooling like dnf5, enhanced security, and broad support across Azure VMs, AKS, and WSL, signaling Microsoft's evolution into a major Linux distribution contributor.

actions/runner-images provides the source and definitions for VM images powering GitHub Actions hosted runners and Microsoft-hosted Azure Pipelines agents. It standardizes Ubuntu, Windows, and macOS environments across x64 and Arm64, preinstalls common development tools, and manages weekly updates, version labels, beta/GA releases, deprecation, and software support policies.

GitHub is resolving degraded availability for Actions Jobs on Hosted Runners in the East US region. While Standard Hosted Runners are showing signs of recovery following mitigation efforts, users with Private Networking remain impacted as Azure works to restore capacity. Users are advised to fail over to alternative regions to mitigate ongoing queue delays and job failures.

On 3 September 2026, ChatGPT, Claude, and Grok experienced overlapping outages, while Gemini reported no official incident. Although Azure issues fueled speculation about shared infrastructure, no common cause was confirmed. The disruptions highlighted AI’s dependence on concentrated cloud providers and the growing importance of resilience for business workflows, developer APIs, and government systems.