Configure Kueue job queues and deploy batch workloads on GKE clusters.
npx skills add https://github.com/wangx7/skills-collection --skill gke-batch-hpc
Please help me install this Agent Skill. Skill: gke-batch-hpc Source: https://github.com/wangx7/skills-collection/tree/main/google-skills/skills/cloud/gke-batch-hpc Command: npx skills add https://github.com/wangx7/skills-collection --skill gke-batch-hpc
This Skill eliminates the manual effort and high risk of errors associated with configuring and scaling batch processing and high-performance computing (HPC) workloads on GKE, which is critical for handling large parallel jobs, scientific simulations, and distributed training tasks.
Use the gke-batch-hpc skill to configure a Kueue job queue and deploy a batch data processing workload on your existing GKE cluster.
High-intent search queries and answers about installing and using this skill.
Run batch processing and HPC workloads on GKE by configuring Kubernetes Jobs, JobSets, and Kueue for centralized queuing. This approach supports parallel data pipelines, MPI applications, and ML training while automating resource scaling.
Yes, you can reduce costs for interruptible batch jobs on GKE using Spot VMs. Configure Spot VMs with fallback patterns and scale-to-zero node pools to automatically shut down resources when no jobs are running.
Kueue provides centralized job queuing and fair resource allocation across teams for HPC tasks. It orchestrates multi-job workflows via JobSet, ensuring large-scale parallel data pipelines and distributed training jobs execute efficiently.
Deploy distributed MPI applications on GKE using the Kubeflow MPI Operator. Combine this with low-latency compact node placement to optimize tightly-coupled workloads like computational fluid dynamics simulations.
Yes, GKE supports low-latency compact node placement specifically for tightly-coupled HPC workloads. This configuration minimizes network latency, improving performance for distributed MPI applications and scientific simulations.
Scale GKE resources to zero by configuring scale-to-zero node pools that automatically shut down when no jobs are active. Combine this with Kueue job queues to manage demand and optimize Spot VM costs.
fajshahProvision and manage multi-cloud GKE clusters with kubectl connectivity.
googleAutomate batch and HPC workload deployment on Google Kubernetes Engine.
pluginagentmarketplaceOrchestrates end-to-end ML training pipelines with PyTorch DDP, Optima, and Kubeflow.
hotriluanDeploy cloud infrastructure across Cloudflare, Docker, Google Cloud, and Kubernetes.
