Volume 343, Issue 2 · 11 articles from 2 issues
Part 2 of Red Hat's edge MLOps series covers platforms, accelerators, memory and messaging for edge AI inference.
The second part of Red Hat's MLOps at the edge series follows a signed, optimized OCI model artifact out to edge devices.
It explains why edge AI differs from data center inference, where hardware is standardized and memory plentiful, since edge fleets span heterogeneous, constrained platforms.
The article covers platform selection, hardware accelerator dependency management, shared memory and latency challenges, as well as messaging patterns connecting sensors to inference and inference to downstream systems.
Red Hat explains why it built Red Hat Hardened Images to deliver containers with zero known CVEs.
A Red Hat engineering director uses his own long-running open source app to show how unmaintained container images silently accumulate vulnerabilities.
The National Vulnerability Database recorded 87,487 CVEs in the last 12 months, roughly 240 per day and an 82% increase over the prior year, and packaging a simple utility in a standard Linux image imports hundreds of unused packages.
Red Hat built Red Hat Hardened Images to keep containers secure without hours of manual patching.
Red Hat says the EU Cyber Resilience Act makes secure-by-design practices essential for open source software.
Red Hat notes that the EU Cyber Resilience Act takes full effect December 11, 2027, and its first vulnerability and incident reporting obligations went live September 11, 2026.
Citing Linux Foundation research, it says 66% of companies remain unfamiliar with the CRA and only 41% expect full compliance by the deadline.
With an estimated 97% of code in most codebases coming from open source, the post outlines how organizations can understand their dependencies and prepare.
Cloudera validated Cloudera Data Platform on Red Hat OpenShift Virtualization across more than 100 virtualized nodes.
Red Hat explains how OpenShift lets organizations modernize VMs alongside containers without immediately refactoring applications.
Cloudera has validated Cloudera Data Platform, traditionally deployed on bare-metal servers alongside OpenShift, on Red Hat OpenShift Virtualization at scale across more than 100 virtualized nodes.
This lets enterprises consolidate application and data workloads on a single platform, and the post details how the architecture was validated.
Red Hat shows how on-premise Lightspeed in Satellite helps triage RHEL vulnerabilities and apply remediations.
Red Hat Lightspeed can run on premise so system data never leaves the customer network, and this walkthrough follows an earlier guide on installing it that way.
It covers using Lightspeed recommendations inside Red Hat Satellite to manage RHEL fleet health, including a 'total risk' visualization of potential issues across the fleet.
The post also explains how to identify and triage vulnerabilities by business risk and apply proactive Advisor remediations efficiently.
Red Hat picks five OpenShift operators, starting with GitOps, that show what the platform can do.
Red Hat compares OpenShift operators to a killer app for the platform.
It explains that some operators install and manage applications such as databases, while others add new concepts like GitOps, CI/CD pipelines or certificate management through custom resource definitions.
The post selects five of the most important operators, beginning with GitOps, and describes how each extends what OpenShift clusters can do.
Red Hat's Q3 2026 open source security report looks beyond vulnerability counts to practical exposure.
Red Hat's quarterly Open Source Security Signals report examines open source security for Q3 2026, aiming to give a fuller picture of the quarter's signals.
It argues that counting disclosed or critical vulnerabilities is not enough, because a critical flaw may pose little real exposure while a lower-severity weakness can become serious when chained.
Deep dependency trees, compromised credentials and permissive configurations also shape risk.
Red Hat's MCP gateway is GA in Connectivity Link 1.5, governing AI agent tool traffic on OpenShift.
Following a technology preview, the Model Context Protocol gateway is now generally available as part of Red Hat Connectivity Link 1.5 and integrated into the Red Hat AI portfolio, expanding Red Hat's AgentOps capabilities to govern AI agent tool traffic at scale.
New features include gateway isolation, stateless session support, prompts and resources support, MCP guardrails and security hardening.
Teams can isolate traffic with dedicated gateway instances or a shared gateway with separate components.
Lightwell backported fixes plus Ansible Automation Platform let teams patch app libraries without version upgrades.
Many enterprise applications cannot simply upgrade to the latest library version to fix a vulnerability, and the post cites a 2025 average remediation time of over five months for complex enterprise applications.
Red Hat explains how Lightwell delivers backported security fixes at exact production library versions, with signed artifacts and SBOMs.
Red Hat Ansible Automation Platform then acts as the governed execution layer for safe, auditable remediation at scale.
Red Hat argues enterprises should mix open-weight and frontier models to match cost to each AI use case.
Using an airline cabin analogy, Red Hat argues that most enterprises run every AI request at the same first-class price per token.
A simple task such as checking an account balance and a complex multi-agent fraud investigation should not cost the same.
The post makes the case for balancing open-weight and frontier models according to use case to improve AI return on investment.
Red Hat recaps Q3 2026 network automation updates, including the GA automation orchestrator for Ansible.
Red Hat says Q3 2026 showed network automation has become critical for AI workloads, edge computing and hybrid cloud.
It highlights the general availability of automation orchestrator, an add-on for Ansible Automation Platform 2.7 that combines job templates and automation libraries with logic nodes, event-driven triggers and AI agent recommendations.
The post also covers updates to Ansible Content Collections aimed at making network automation more secure and predictable.