gke-inference

GKE AI/ML Inference

This reference covers deploying AI/ML inference workloads on GKE using Google's Inference Quickstart (GIQ) and best practices for LLM serving.

MCP Tools: apply_k8s_manifest, get_k8s_resource, get_k8s_logs, get_k8s_rollout_status, describe_k8s_resource, list_k8s_events. CLI-only: gcloud container ai profiles *

When to Use

• Deploy an AI model (Llama, Gemma, Mistral, etc.) to GKE
• Generate optimized Kubernetes manifests for inference
• Select GPU/TPU accelerators for model serving
• Configure autoscaling for LLM inference

Prerequisites

• A golden path GKE Autopilot cluster (GPU workloads are supported via ComputeClasses and NAP)
• gcloud CLI authenticated
• Sufficient GPU/TPU quota in the target region

Workflow

1. Discovery: Find Models and Hardware

# List all supported models
gcloud container ai profiles models list --quiet
 
# Find valid accelerator/server combinations for a model
gcloud container ai

2. Generate Manifest

Parameters:

• --model: Model ID (e.g., gemma-2-9b-it, llama-3-8b)
• --model-server: Inference server (vllm, tgi, triton, tensorrt-llm)
• --accelerator-type: GPU/TPU type (nvidia-l4, nvidia-tesla-a100, nvidia-h100-80gb)
• --target-ntpot-milliseconds: Target Normalized Time Per Output Token (optional, for latency optimization)

Example:

3. Review and Deploy

Some models require Hugging Face tokens. Create a Kubernetes Secret and reference it in the manifest.

GPU ComputeClass for Inference

For Autopilot clusters, create a ComputeClass to target GPU nodes:

Accelerator Selection Guide

Autoscaling LLM Inference

GPU-based autoscaling

Use custom metrics for GPU utilization:

Best practices for inference autoscaling

1. Use DCGM metrics: Golden path enables DCGM monitoring for GPU utilization metrics
2. Set appropriate minReplicas: At least 1 for always-on serving; 0 for batch/on-demand
3. Tune scale-down delay: LLM model loading is slow; use longer stabilization windows
4. Consider queue depth: Scale on pending requests rather than pure GPU utilization for latency-sensitive workloads

Optimization Tips

• Quantization: Use quantized models (GPTQ, AWQ) to reduce GPU memory and increase throughput
• Batching: Configure model server batch size for throughput vs latency trade-off
• Tensor parallelism: Split large models across multiple GPUs within a node
• KV cache optimization: Tune --gpu-memory-utilization in vLLM for KV cache allocation

Troubleshooting