9 documented GCP GKE cost inefficiencies, each with how to detect it and how to fix it.
GCP GKECER-0287
Workloads are constrained to run only on Spot-based capacity with no viable path to standard nodes when Spot capacity is reclaimed or unavailable. While Spot reduces unit cost, rigid dependence can create hidden costs by requiring standby...
GCP GKECER-0193
As environments scale, GKE clusters tend to accumulate artifacts from ephemeral workloads, dev environments, or incomplete job execution. PVCs can continue to retain Persistent Disks, Services may continue to expose public IPs and...
GCP GKECER-0270
In GKE environments, it is common for unused Kubernetes resources to accumulate over time. Examples include Persistent Volume Claims (PVCs) that retain provisioned Persistent Disks, or Services of type LoadBalancer that continue to front...
GCP GKECER-0498
The GKE cluster autoscaler has autoscaling profiles that decide how eagerly it removes nodes. Standard clusters default to the balanced profile, which Google describes as prioritizing keeping more resources readily available for incoming...
GCP GKECER-0499
Batch jobs, CI runners, rendering, data processing and other restart-tolerant workloads on GKE frequently run on standard node pools or standard Autopilot Pods at regular prices. GKE supports Spot VMs in Standard node pools and Spot Pods...
GCP GKECER-0501
Kubernetes Pods request GPUs in whole numbers, so by default an entire physical GPU is allocated to one container even when that container needs only a fraction of it. Small inference services, notebooks, development environments and light...
GCP GKECER-0560
GKE supports each minor version for 14 months of standard support. Clusters enrolled in the Extended release channel can stay on a minor version for up to 24 months in total, receiving security patches for roughly 10 more months during the...
GCP GKECER-0562
GKE clusters created for testing, demos, proofs of concept, migrations or short-lived projects are often left running after the work ends. Even with no customer workloads, every cluster pays the flat hourly cluster management fee, and a...
GCP GKECER-0573
GKE Autopilot bills general-purpose Pods on the CPU, memory and ephemeral storage they request, not on what they use, and not on the nodes underneath. Every vCPU and GiB requested above real need is billed for as long as the Pod runs....
