Receive daily AI-curated summaries of engineering articles from top tech companies worldwide.
Endigest AI Core Summary
Kubernetes v1.35 introduces Extended Toleration Operators (alpha) that add Gt (Greater Than) and Lt (Less Than) operators to pod tolerations, enabling numeric threshold-based scheduling decisions.
•Previously, tolerations only supported Equal and Exists operators, forcing workarounds like discrete taint categories or external admission controllers for numeric comparisons.
•The new Gt and Lt operators allow workloads to specify thresholds such as 'schedule only on nodes with failure probability less than 5%' or 'require GPU compute score greater than 800'.
•Unlike NodeAffinity, taints with numeric operators preserve eviction semantics (NoExecute + tolerationSeconds) and provide centralized, node-side policy control.
•Real-world use cases include spot instance SLA protection, GPU tier-based AI workload placement, cost-optimized batch scheduling, and disk IOPS performance guarantees.
•The feature is enabled via the TaintTolerationComparisonOperators feature gate and requires numeric
This summary was automatically generated by AI based on the original article and may not be fully accurate.
Kubernetes
Google Cloud
AWS