Google Cloud Blog
Follow
Unlock next-gen VMs using GKE compute classes and Compute Flexible CUDs
Organizations seek to adopt new Google Cloud machine series, like N4 and C4, for improved performance and cost-efficiency. Migrating to new hardware presents challenges such as compatibility testing, regional capacity issues, and financial commitment complexities. Google Cloud offers a solution by combining GKE compute classes with Compute Flexible Committed Use Discounts (Flex CUDs). GKE compute classes allow workloads to be assigned a prioritized list of machine families for autoscaling. This means GKE can attempt to use new hardware first, falling back to older, established options if necessary. This phased rollout minimizes operational risk and downtime, as demonstrated by Shopify's successful adoption. For demanding workloads, compute classes can prioritize newer C-series VMs like C4 or C4D, with fallbacks to older generations. Flex CUDs provide financial adaptability, applying discounts to total eligible compute spend across various machine families, unlike traditional resource-based CUDs. This ensures that discounts automatically follow workloads, even as they switch between machine types due to GKE's fallback mechanisms. Verve Group utilizes this flexibility, leveraging Flex CUDs across multiple machine series. Together, GKE compute classes and Flex CUDs enable safe innovation, cost optimization, enhanced resilience, and simplified operations when adopting new hardware. This integrated approach allows businesses to confidently leverage advancements in Google Cloud's compute landscape.