Google Cloud · Infrastructure · YoctoIT tech page

Compute Engine

Google's VMs: per-second billing, automatic sustained-use discounts and transparent live migration — IaaS without surprises.

FOCUS · VMS PAID RIGHTAutomatic discounts, custom sizes and invisible maintenance: well-mannered IaaS
Request the Free Assessment Learn more
YoctoIT material for clients and partners · Google Cloud, BigQuery, Gemini and the other products mentioned are trademarks of Google LLC.
01 · What it is

Compute Engine, made clear.

Compute Engine has some unique good manners: sustained-use discounts apply BY THEMSELVES (nothing to reserve), sizes are built to measure (independent vCPUs and RAM), and Google's live migration moves VMs during maintenance without reboots. IaaS with fewer traps than usual.

Al secondo
the billing: a VM on for 10 minutes costs 10 minutes
SUD
Sustained Use Discounts: the automatic discount up to 30%, no commitments
Live migrate
host maintenance without reboots: the VM doesn't notice
Compute Engine
OFFICIAL GOOGLE CLOUD BRANDING · COMPUTE ENGINE
CONSOLE REALE · COMPUTE ENGINE SSH · FONTE: GOOGLE CLOUD DOCS
REAL CONSOLE · COMPUTE ENGINE SSH · SOURCE: GOOGLE CLOUD DOCS
02 · How to use it well

The things that make the difference.

Google's IaaS

Applicazioni Windows & Linuxthe usual workloads
Custom sizes
Spot VM
MIGs · managed groups
measure · savings · scale
Automatic discounts + CUDsthe price that optimizes itself
Google's global network underneaththe extra gear
Normal VMs, better economics

Custom machine type

The workload's exact size, not the nearest in the catalog: waste zeroed by design.

Spot for batch

Up to -91% on interruptible workloads: rendering, tests, queues — organized savings.

MIGs and autohealing

Instance groups with health checks and automatic replacement: infrastructure that heals itself.

CUDs where they pay

Committed use on the stable workloads: the spot/on-demand/committed mix is a FinOps project.

03 · In depth

Families, disks and the discount models

Compute Engine thinks in families (economical E2, N4/C4 general purpose on Titanium, C3 performance, M3 memory, custom machine types for the exact size) with native live migration: host maintenance doesn't stop the VMs. Hyperdisk decouples IOPS and throughput from capacity; discounts are automatic (sustained use) or committed (flexible or resource CUDs); Spot VMs cut up to 91% for interruptible workloads.

  • Custom machine type — made-to-measure vCPUs and RAM: the exact size, not the nearest
  • Live migration — Google's invisible maintenance: the VM doesn't reboot
  • Hyperdisk — independently tuned IOPS/throughput: the disk tuned to the workload
  • CUD flessibili — the spend commitment covering different families and regions
  • Sustained use — the automatic discount growing with use: zero planning
  • Spot VM — up to -91% for batch and CI: the interruptible that pays
04 · Numbers and lifecycle

The numbers that matter.

99,95%
SLA for VMs across zones
-91%
the maximum Spot discount
0
maintenance reboots: live migration as standard
s
the provisioning: the VM ready in seconds
The right compute is family+disk+discount: right-sizing with Recommender and CUDs tuned to consumption — we keep the bill in check.
05 · Use cases

Where it really pays off.

Lift & shift

The servers migrating fast, onto a clean IaaS.

Ephemeral environments

Dev, test and trials switched on by consumption.

Certified SAP

The sizes certified for HANA and NetWeaver: here too.

IaaS is judged on the invoice: Compute Engine's, well governed, is a pleasure to read.