14 documented GCP Compute Engine cost inefficiencies, each with how to detect it and how to fix it.
GCP Compute EngineCER-0282
Workloads with predictable, long-running compute usage continue to run entirely on on-demand pricing instead of leveraging Committed Use Discounts. For stable environments, such as production services or continuously running batch...
GCP Compute EngineCER-0089
VM-based Committed Use Discounts in GCP offer cost savings for predictable workloads, but they are rigid: they apply only to specified VM types, quantities, and regions. When organizations evolve their architecture - such as moving to GKE...
GCP Compute EngineCER-0095
Development and test environments on Compute Engine are commonly provisioned and left running around the clock, even if only used during business hours. This results in wasteful spend on compute time that could be eliminated by scheduling...
GCP Compute EngineCER-0048
GCP VM instances are often provisioned with more CPU or memory than needed, especially when using custom machine types or legacy templates. If an instance consistently consumes only a small portion of its allocated resources, it likely...
GCP Compute EngineCER-0460
An idle VM is a running Compute Engine instance that does no useful work: its CPU and network activity come mostly from the OS and monitoring agents. Typical sources are VMs left over from finished projects, migrations, proofs of concept,...
GCP Compute EngineCER-0461
Persistent Disk and Hyperdisk volumes are billed for their full provisioned size until they are deleted, whether or not any VM is using them. Disks become detached and forgotten in predictable ways: a VM is deleted with auto-delete turned...
GCP Compute EngineCER-0462
A reserved static external IPv4 address that is not assigned to any resource keeps billing every hour, and Google charges a higher rate for a reserved-but-unused address than for an address in use. Addresses end up unused when the VM they...
GCP Compute EngineCER-0463
Stopping a Compute Engine VM ends vCPU and memory charges, which is why stopping is the usual response to an unneeded machine. But the VM's resources remain: its boot and data disks keep billing for their full provisioned size, a static...
GCP Compute EngineCER-0464
Compute Engine on-demand reservations guarantee capacity for a number of VMs of a given shape in a zone. Google charges for the reserved resources at the same on-demand rate as running VMs, including any applicable discounts, for as long...
GCP Compute EngineCER-0465
Custom images are billed for storage for as long as they exist, whether or not any VM or disk is ever created from them again. Image build pipelines (Packer, CI jobs, golden-image processes) typically publish a new image version for every...
GCP Compute EngineCER-0466
Standard and archive snapshots of Compute Engine disks are not deleted when their source disk is deleted, and they keep billing for stored data until someone deletes them. Snapshot schedules make this worse by default: a schedule's...
GCP Compute EngineCER-0467
Hyperdisk separates performance from capacity: Hyperdisk Balanced bills provisioned IOPS and throughput above a free baseline, Hyperdisk Extreme bills provisioned IOPS and Hyperdisk Throughput bills provisioned throughput. Those charges...
GCP Compute EngineCER-0468
Spot VMs run the same machine types as standard VMs at discounts of up to 91% for many machine types, GPUs, TPUs and Local SSDs, in exchange for no availability guarantee: Compute Engine can preempt them at any time. Batch processing, CI...
GCP Compute EngineCER-0469
N1 is Compute Engine's first-generation general-purpose series, running on Intel platforms from Ivy Bridge to Skylake. Many long-lived VMs, instance templates and GKE node pools still use n1-* machine types because they were created years...
