These are all the cool new GKE releases/announcements from May. K8s 1.33 released on Rapid Channel. A new joint collaboration between Google and a number of companies to improve llm serving was announced. New Performance HPA released with up to 3x performance improvements. Confidential node support for GPU vms. CoC is now the default on autopilot on 1.32+ clusters. Container thread detection added support for new threats. TPU support in vllm is now GA. Network intelligence center added support for GKE IP Masquerade analyzer. Learning GKE for AI/ML is easier than ever with GKE AI Labs.
Chapters:
0:00 – Welcome to GKE May Edition
0:12 – Kubernetes 1.33 in rapid channel
0:23- Introducing llm-d
0:44- Performance HPA is Available
1:01- Confidential node and GPU support
01:35- COC is now defaulting GKE
1:45- Container threat detection added support for new threats
1:59 – TPU is now GA
2:06- Integrated a new feature in network analyzer
2:27 – New platform for Kubernetes tutorials
Resources:
Learn more →https://goo.gle/confidential-gpu-nodes
Learn more →https://goo.gle/gke-vllm-tpu
Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech
#GoogleCloud #GenerativeAI
Speakers: Mofi Rahman
Products Mentioned: Google Kubernetes Engine (GKE)
