# Compute Engine and Managed Instance Groups — Google Cloud Platform

Source: https://www.skillbyai.com/en/gcp/c-gce

> Choose machine types and run resilient VM fleets.

## Virtual machines and fleets of them

**Compute Engine** provides VMs with your choice of OS. **Machine families** target different needs: **E2** for cost-efficient general purpose, **N2/N4/C3/C4** for balanced or high-performance general purpose, compute-optimised, memory-optimised and accelerator (GPU/TPU) families for specialised work, and **custom machine types** let you pick vCPU and memory exactly. Disks are **Persistent Disk** or **Hyperdisk** block storage, zonal or regional. For fleets, write an **instance template** and create a **managed instance group (MIG)**: it keeps a target number of identical VMs, **autoheals** them using health checks, **autoscales** on CPU, load-balancer utilisation or custom metrics, and performs **rolling updates** to a new template. A **regional MIG** spreads VMs across zones. **Spot VMs** offer deep discounts for interruptible work but can be stopped whenever Compute Engine needs the capacity.

## A regional managed instance group

One template, many identical VMs across zones, kept healthy and sized by the group.

![A template card on the left feeding three zone columns of identical server blocks, with a heart-shaped health check and up-down scaling arrows.](assets/figures/gcp/section-3-map.svg) — Figure 3.1 — A regional MIG built from an instance template.

## Template, regional MIG and autoscaling

The group recreates unhealthy VMs and adds VMs when CPU rises.

```bash
gcloud compute instance-templates create web-tpl-v1 \
  --machine-type=e2-medium --image-family=debian-12 --image-project=debian-cloud \
  --metadata-from-file=startup-script=startup.sh --tags=web

gcloud compute instance-groups managed create web-mig \
  --template=web-tpl-v1 --size=2 --region=asia-south1

gcloud compute instance-groups managed set-autoscaling web-mig --region=asia-south1 \
  --min-num-replicas=2 --max-num-replicas=8 --target-cpu-utilization=0.65
```

## Treat VMs as cattle

Configure VMs from the template and a startup script or image, not by SSHing in and editing files. A MIG will recreate VMs during autohealing and updates, and hand-made changes disappear.

**Quiz:** What does a managed instance group do when a VM fails its health check?

- [x] Recreates the VM from the instance template
- [ ] Sends an email and waits
- [ ] Deletes the whole group
- [ ] Converts it to a Spot VM

*Answer:* Recreates the VM from the instance template. Autohealing recreates unhealthy instances from the template.